DHPT Assignment Notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Mechanical Part of Speech Tagging for the Latin Language\n",
    "## Digital Humanities: Practice & Theory\n",
    "## Fergus J. P. Walsh\n",
    "\n",
    "## Companion Jupyter Notebook"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Text preparation and tagset conversion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Functions to convert the Universal Dependencies tagset to Perseus-style tags. Note how some tags must be assigned by lemma; Perseus classes \"nam\", \"enim\", \"donec\" etc. as conjunctions, but in the LLCT, for example, they are adverbs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "def LLCTtoPerseus(LLCTtag, LLCTlemma):\n",
    "    noun = re.search(\"NOUN|PROPN\", LLCTtag)\n",
    "    verb = re.search(\"VERB|AUX\", LLCTtag)\n",
    "    adjective = re.search(\"ADJ|DET|unus|duo\", LLCTtag)\n",
    "    adverb = re.search(\"ADV|PART|non\", LLCTtag)\n",
    "    conjunction = re.search(\"SCONJ|CCONJ|nam|enim|donec|ac|quoque|igitur\", LLCTtag)\n",
    "    adposition = re.search(\"ADP\", LLCTtag)\n",
    "    pronoun = re.search(\"PRON|qui|quidam|hic|iste|ipse\", LLCTtag)\n",
    "    numeral = re.search(\"NUM\", LLCTtag)\n",
    "    interjection = re.search(\"INTJ\", LLCTtag)\n",
    "    punctuation = re.search(\"PUNCT\", LLCTtag)\n",
    "    if noun:\n",
    "        pos = \"n\"\n",
    "    elif verb:\n",
    "        pos = \"v\"\n",
    "    elif adjective:\n",
    "        pos = \"a\"\n",
    "    elif adverb:\n",
    "        pos = \"d\"\n",
    "    elif conjunction:\n",
    "        pos = \"c\"\n",
    "    elif adposition:\n",
    "        pos = \"r\"\n",
    "    elif pronoun:\n",
    "        pos = \"p\"\n",
    "    elif numeral:\n",
    "        pos = \"m\"\n",
    "    elif interjection:\n",
    "        pos = \"i\"\n",
    "    elif punctuation:\n",
    "        pos = \"u\"\n",
    "    else:\n",
    "        pos = \"e\"\n",
    "    \n",
    "    first = re.search(\"'Person': '1'\", LLCTtag)\n",
    "    second = re.search(\"'Person': '2'\", LLCTtag)\n",
    "    third = re.search(\"'Person': '3'\", LLCTtag)\n",
    "    if first:\n",
    "        person = \"1\"\n",
    "    elif second:\n",
    "        person = \"2\"\n",
    "    elif third:\n",
    "        person = \"3\"\n",
    "    else:\n",
    "        person = \"-\"\n",
    "        \n",
    "    sing = re.search(\"'Number': 'Sing'\", LLCTtag)\n",
    "    plur = re.search(\"'Number': 'Plur'\", LLCTtag)\n",
    "    if sing:\n",
    "        number = \"s\"\n",
    "    elif plur:\n",
    "        number = \"p\"\n",
    "    else:\n",
    "        number = \"-\"\n",
    "        \n",
    "    present = re.search(\"'Tense': 'Pres'\", LLCTtag)\n",
    "    imperfect = re.search(\"'Aspect': 'Imp'.*('Tense': 'Past')\", LLCTtag)\n",
    "    perfect = re.search(\"'Aspect': 'Perf'.*('Tense': 'Past')\", LLCTtag)\n",
    "    pluperfect = re.search(\"'Tense': 'Pqp'\", LLCTtag)\n",
    "    futureperfect = re.search(\"'Aspect': 'Perf'.*('Tense': 'Fut')\", LLCTtag)\n",
    "    future = re.search(\"'Tense': 'Fut'\", LLCTtag)\n",
    "    if present:\n",
    "        tense = \"p\"\n",
    "    elif imperfect:\n",
    "        tense = \"i\"\n",
    "    elif perfect:\n",
    "        tense = \"r\"\n",
    "    elif pluperfect:\n",
    "        tense = \"l\"\n",
    "    elif futureperfect:\n",
    "        tense = \"t\"\n",
    "    elif future:\n",
    "        tense = \"f\"\n",
    "    else:\n",
    "        tense = \"-\"\n",
    "    \n",
    "    indicative = re.search(\"'Mood': 'Ind'\", LLCTtag)\n",
    "    subjunctive = re.search(\"'Mood': 'Sub'\", LLCTtag)\n",
    "    infinitive = re.search(\"'VerbForm': 'Inf'\", LLCTtag)\n",
    "    imperative = re.search(\"'Mood': 'Imp'\", LLCTtag)\n",
    "    participle = re.search(\"'VerbForm': 'Part'\", LLCTtag)\n",
    "    gerund = re.search(\"'VerbForm': 'Ger'\", LLCTtag)\n",
    "    gerundive = re.search(\"'VerbForm': 'Gdv'\", LLCTtag)\n",
    "    if indicative:\n",
    "        mood = \"i\"\n",
    "    elif subjunctive:\n",
    "        mood = \"s\"\n",
    "    elif infinitive:\n",
    "        mood = \"n\"\n",
    "    elif imperative:\n",
    "        mood = \"m\"\n",
    "    elif imperative:\n",
    "        mood = \"m\"\n",
    "    elif participle:\n",
    "        mood = \"p\"\n",
    "    elif gerund:\n",
    "        mood = \"d\"\n",
    "    elif gerundive:\n",
    "        mood = \"g\"\n",
    "    else:\n",
    "        mood = \"-\"\n",
    "    \n",
    "    active = re.search(\"'Voice': 'Act'\", LLCTtag)\n",
    "    deponent = LLCTlemma in deponent_list_iu\n",
    "    passive = not deponent and re.search(\"'Voice': 'Pass'\", LLCTtag)\n",
    "    if active:\n",
    "        voice = \"a\"\n",
    "    elif passive:\n",
    "        voice = \"p\"\n",
    "    elif deponent:\n",
    "        voice = \"d\"\n",
    "    else:\n",
    "        voice = \"-\"\n",
    "    \n",
    "    masculine = re.search(\"'Gender': 'Masc'\", LLCTtag)\n",
    "    feminine = re.search(\"'Gender': 'Fem'\", LLCTtag)\n",
    "    neuter = re.search(\"'Gender': 'Neut'\", LLCTtag)\n",
    "    if masculine:\n",
    "        gender = \"m\"\n",
    "    elif feminine:\n",
    "        gender = \"f\"\n",
    "    elif neuter:\n",
    "        gender = \"n\"\n",
    "    else:\n",
    "        gender = \"-\"\n",
    "        \n",
    "    nominative = re.search(\"'Case': 'Nom'\", LLCTtag)\n",
    "    genitive = re.search(\"'Case': 'Gen'\", LLCTtag)\n",
    "    dative = re.search(\"'Case': 'Dat'\", LLCTtag)\n",
    "    accusative = re.search(\"'Case': 'Acc'\", LLCTtag)\n",
    "    vocative = re.search(\"'Case': 'Voc'\", LLCTtag)\n",
    "    ablative = re.search(\"'Case': 'Abl'\", LLCTtag)\n",
    "    locative = re.search(\"'Case': 'Loc'\", LLCTtag)\n",
    "    if nominative:\n",
    "        case = \"n\"\n",
    "    elif genitive:\n",
    "        case = \"g\"\n",
    "    elif dative:\n",
    "        case = \"d\"\n",
    "    elif accusative:\n",
    "        case = \"a\"\n",
    "    elif vocative:\n",
    "        case = \"v\"\n",
    "    elif ablative:\n",
    "        case = \"b\"\n",
    "    elif locative:\n",
    "        case = \"l\"\n",
    "    else:\n",
    "        case = \"-\"\n",
    "    \n",
    "    positive = re.search(\"'Degree': 'Pos'\", LLCTtag)\n",
    "    comparative = re.search(\"'Degree': 'Cmp'\", LLCTtag)\n",
    "    superlative = re.search(\"'Degree': 'Sup'\", LLCTtag)\n",
    "    if comparative:\n",
    "        degree = \"c\"\n",
    "    elif superlative:\n",
    "        degree = \"s\"\n",
    "    elif positive:\n",
    "        degree = \"p\"\n",
    "    else:\n",
    "        degree = \"-\"\n",
    "\n",
    "    LLCT_new_tag = f\"{pos}.{person}.{number}.{tense}.{mood}.{voice}.{gender}.{case}.{degree}\"\n",
    "    return LLCT_new_tag\n",
    "\n",
    "def PROIELtoPerseus(PROIELtag, PROIELlemma):\n",
    "    noun = re.search(\"NOUN|PROPN\", PROIELtag)\n",
    "    verb = re.search(\"VERB\", PROIELtag)\n",
    "    adjective = re.search(\"ADJ|DET|unus|duo\", PROIELtag)\n",
    "    adverb = re.search(\"ADV\", PROIELtag)\n",
    "    conjunction = re.search(\"SCONJ|CCONJ|nam|enim|donec|ac|quoque|igitur\", PROIELtag)\n",
    "    adposition = re.search(\"ADP\", PROIELtag)\n",
    "    pronoun = re.search(\"PRON|qui|quidam|hic|iste|ipse|ille\", PROIELtag)\n",
    "    numeral = re.search(\"NUM\", PROIELtag)\n",
    "    interjection = re.search(\"INTJ\", PROIELtag)\n",
    "    punctuation = re.search(\"PUNCT\", PROIELtag)\n",
    "    if noun:\n",
    "        pos = \"n\"\n",
    "    elif verb:\n",
    "        pos = \"v\"\n",
    "    elif adjective:\n",
    "        pos = \"a\"\n",
    "    elif adverb:\n",
    "        pos = \"d\"\n",
    "    elif conjunction:\n",
    "        pos = \"c\"\n",
    "    elif adposition:\n",
    "        pos = \"r\"\n",
    "    elif pronoun:\n",
    "        pos = \"p\"\n",
    "    elif numeral:\n",
    "        pos = \"m\"\n",
    "    elif interjection:\n",
    "        pos = \"i\"\n",
    "    elif punctuation:\n",
    "        pos = \"u\"\n",
    "    else:\n",
    "        pos = \"n\"\n",
    "    \n",
    "    first = re.search(\"'Person': '1'\", PROIELtag)\n",
    "    second = re.search(\"'Person': '2'\", PROIELtag)\n",
    "    third = re.search(\"'Person': '3'\", PROIELtag)\n",
    "    if first:\n",
    "        person = \"1\"\n",
    "    elif second:\n",
    "        person = \"2\"\n",
    "    elif third:\n",
    "        person = \"3\"\n",
    "    else:\n",
    "        person = \"-\"\n",
    "        \n",
    "    sing = re.search(\"'Number': 'Sing'\", PROIELtag)\n",
    "    plur = re.search(\"'Number': 'Plur'\", PROIELtag)\n",
    "    if sing:\n",
    "        number = \"s\"\n",
    "    elif plur:\n",
    "        number = \"p\"\n",
    "    else:\n",
    "        number = \"-\"\n",
    "        \n",
    "    present = re.search(\"'Tense': 'Pres'\", PROIELtag)\n",
    "    imperfect = re.search(\"'Aspect': 'Imp'.*('Tense': 'Past')\", PROIELtag)\n",
    "    perfect = re.search(\"'Aspect': 'Perf'.*('Tense': 'Past')\", PROIELtag)\n",
    "    pluperfect = re.search(\"'Tense': 'Pqp'\", PROIELtag)\n",
    "    futureperfect = re.search(\"'Aspect': 'Perf'.*('Tense': 'Fut')\", PROIELtag)\n",
    "    future = re.search(\"'Tense': 'Fut'\", PROIELtag)\n",
    "    if present:\n",
    "        tense = \"p\"\n",
    "    elif imperfect:\n",
    "        tense = \"i\"\n",
    "    elif perfect:\n",
    "        tense = \"r\"\n",
    "    elif pluperfect:\n",
    "        tense = \"l\"\n",
    "    elif futureperfect:\n",
    "        tense = \"t\"\n",
    "    elif future:\n",
    "        tense = \"f\"\n",
    "    else:\n",
    "        tense = \"-\"\n",
    "    \n",
    "    indicative = re.search(\"'Mood': 'Ind'\", PROIELtag)\n",
    "    subjunctive = re.search(\"'Mood': 'Sub'\", PROIELtag)\n",
    "    infinitive = re.search(\"'VerbForm': 'Inf'\", PROIELtag)\n",
    "    imperative = re.search(\"'Mood': 'Imp'\", PROIELtag)\n",
    "    participle = re.search(\"'VerbForm': 'Part'\", PROIELtag)\n",
    "    gerund = re.search(\"'VerbForm': 'Ger'\", PROIELtag)\n",
    "    gerundive = re.search(\"'VerbForm': 'Gdv'\", PROIELtag)\n",
    "    if indicative:\n",
    "        mood = \"i\"\n",
    "    elif subjunctive:\n",
    "        mood = \"s\"\n",
    "    elif infinitive:\n",
    "        mood = \"n\"\n",
    "    elif imperative:\n",
    "        mood = \"m\"\n",
    "    elif imperative:\n",
    "        mood = \"m\"\n",
    "    elif participle:\n",
    "        mood = \"p\"\n",
    "    elif gerund:\n",
    "        mood = \"d\"\n",
    "    elif gerundive:\n",
    "        mood = \"g\"\n",
    "    else:\n",
    "        mood = \"-\"\n",
    "    \n",
    "    active = re.search(\"'Voice': 'Act'\", PROIELtag)\n",
    "    deponent = PROIELlemma in deponent_list_iu\n",
    "    passive = not deponent and re.search(\"'Voice': 'Pass'\", PROIELtag)\n",
    "    if active:\n",
    "        voice = \"a\"\n",
    "    elif passive:\n",
    "        voice = \"p\"\n",
    "    elif deponent:\n",
    "        voice = \"d\"\n",
    "    else:\n",
    "        voice = \"-\"\n",
    "    \n",
    "    masculine = re.search(\"'Gender': 'Masc'\", PROIELtag)\n",
    "    feminine = re.search(\"'Gender': 'Fem'\", PROIELtag)\n",
    "    neuter = re.search(\"'Gender': 'Neut'\", PROIELtag)\n",
    "    if masculine:\n",
    "        gender = \"m\"\n",
    "    elif feminine:\n",
    "        gender = \"f\"\n",
    "    elif neuter:\n",
    "        gender = \"n\"\n",
    "    else:\n",
    "        gender = \"-\"\n",
    "        \n",
    "    nominative = re.search(\"'Case': 'Nom'\", PROIELtag)\n",
    "    genitive = re.search(\"'Case': 'Gen'\", PROIELtag)\n",
    "    dative = re.search(\"'Case': 'Dat'\", PROIELtag)\n",
    "    accusative = re.search(\"'Case': 'Acc'\", PROIELtag)\n",
    "    vocative = re.search(\"'Case': 'Voc'\", PROIELtag)\n",
    "    ablative = re.search(\"'Case': 'Abl'\", PROIELtag)\n",
    "    if nominative:\n",
    "        case = \"n\"\n",
    "    elif genitive:\n",
    "        case = \"g\"\n",
    "    elif dative:\n",
    "        case = \"d\"\n",
    "    elif accusative:\n",
    "        case = \"a\"\n",
    "    elif vocative:\n",
    "        case = \"v\"\n",
    "    elif ablative:\n",
    "        case = \"b\"\n",
    "    else:\n",
    "        case = \"-\"\n",
    "    \n",
    "    positive = re.search(\"'Degree': 'Pos'\", PROIELtag)\n",
    "    comparative = re.search(\"'Degree': 'Cmp'\", PROIELtag)\n",
    "    superlative = re.search(\"'Degree': 'Sup'\", PROIELtag)\n",
    "    if comparative:\n",
    "        degree = \"c\"\n",
    "    elif superlative:\n",
    "        degree = \"s\"\n",
    "    elif positive:\n",
    "        degree = \"p\"\n",
    "    else:\n",
    "        degree = \"-\"\n",
    "\n",
    "    PROIEL_new_tag = f\"{pos}.{person}.{number}.{tense}.{mood}.{voice}.{gender}.{case}.{degree}\"\n",
    "    return PROIEL_new_tag\n",
    "\n",
    "def ITtoPerseus(ITtag, ITlemma):\n",
    "    noun = re.search(\"NOUN\", ITtag)\n",
    "    verb = re.search(\"VERB|[A-S]3\", ITtag)\n",
    "    adjective = re.search(\"ADJ|DET|unus|duo\", ITtag)\n",
    "    adverb = re.search(\"ADV|non\", ITtag)\n",
    "    conjunction = re.search(\"SCONJ|CCONJ|nam|enim|donec|ac|quoque|igitur\", ITtag)\n",
    "    adposition = re.search(\"ADP\", ITtag)\n",
    "    pronoun = re.search(\"PRON|qui|quidam|hic|iste\", ITtag)\n",
    "    numeral = re.search(\"NUM\", ITtag)\n",
    "    interjection = re.search(\"INTJ\", ITtag)\n",
    "    punctuation = re.search(\"PUNCT\", ITtag)\n",
    "    if noun:\n",
    "        pos = \"n\"\n",
    "    elif verb:\n",
    "        pos = \"v\"\n",
    "    elif adjective:\n",
    "        pos = \"a\"\n",
    "    elif adverb:\n",
    "        pos = \"d\"\n",
    "    elif conjunction:\n",
    "        pos = \"c\"\n",
    "    elif adposition:\n",
    "        pos = \"r\"\n",
    "    elif pronoun:\n",
    "        pos = \"p\"\n",
    "    elif numeral:\n",
    "        pos = \"m\"\n",
    "    elif interjection:\n",
    "        pos = \"i\"\n",
    "    elif punctuation:\n",
    "        pos = \"u\"\n",
    "    else:\n",
    "        pos = \"e\"\n",
    "    \n",
    "    first = re.search(\"gen4|gen7\", ITtag)\n",
    "    second = re.search(\"gen5|gen8\", ITtag)\n",
    "    third = re.search(\"gen6|gen9\", ITtag)\n",
    "    if first:\n",
    "        person = \"1\"\n",
    "    elif second:\n",
    "        person = \"2\"\n",
    "    elif third:\n",
    "        person = \"3\"\n",
    "    else:\n",
    "        person = \"-\"\n",
    "        \n",
    "    sing = re.search(\"gen4|gen5|gen6|cas[A-H]\", ITtag)\n",
    "    plur = re.search(\"gen7|gen8|gen9|cas[J-O]\", ITtag)\n",
    "    if sing:\n",
    "        number = \"s\"\n",
    "    elif plur:\n",
    "        number = \"p\"\n",
    "    else:\n",
    "        number = \"-\"\n",
    "        \n",
    "    present = re.search(\"tem1\", ITtag)\n",
    "    imperfect = re.search(\"tem2\", ITtag)\n",
    "    perfect = re.search(\"tem4\", ITtag)\n",
    "    pluperfect = re.search(\"tem5\", ITtag)\n",
    "    futureperfect = re.search(\"tem6\", ITtag)\n",
    "    future = re.search(\"tem3\", ITtag)\n",
    "    if present:\n",
    "        tense = \"p\"\n",
    "    elif imperfect:\n",
    "        tense = \"i\"\n",
    "    elif perfect:\n",
    "        tense = \"r\"\n",
    "    elif pluperfect:\n",
    "        tense = \"l\"\n",
    "    elif futureperfect:\n",
    "        tense = \"t\"\n",
    "    elif future:\n",
    "        tense = \"f\"\n",
    "    else:\n",
    "        tense = \"-\"\n",
    "    \n",
    "    indicative = re.search(\"modA|modJ\", ITtag)\n",
    "    subjunctive = re.search(\"modB|modK\", ITtag)\n",
    "    infinitive = re.search(\"modH|modQ\", ITtag)\n",
    "    imperative = re.search(\"modC|modL\", ITtag)\n",
    "    participle = re.search(\"modD|modM\", ITtag)\n",
    "    gerund = re.search(\"modE|modN\", ITtag)\n",
    "    gerundive = re.search(\"modO\", ITtag)\n",
    "    if indicative:\n",
    "        mood = \"i\"\n",
    "    elif subjunctive:\n",
    "        mood = \"s\"\n",
    "    elif infinitive:\n",
    "        mood = \"n\"\n",
    "    elif imperative:\n",
    "        mood = \"m\"\n",
    "    elif imperative:\n",
    "        mood = \"m\"\n",
    "    elif participle:\n",
    "        mood = \"p\"\n",
    "    elif gerund:\n",
    "        mood = \"d\"\n",
    "    elif gerundive:\n",
    "        mood = \"g\"\n",
    "    else:\n",
    "        mood = \"-\"\n",
    "    \n",
    "    active = re.search(\"mod[A-E]|modG|modH\", ITtag)\n",
    "    deponent = ITlemma in deponent_list_iu\n",
    "    passive = not deponent and re.search(\"mod[J-Q]\", ITtag)\n",
    "    if active:\n",
    "        voice = \"a\"\n",
    "    elif passive:\n",
    "        voice = \"p\"\n",
    "    elif deponent:\n",
    "        voice = \"d\"\n",
    "    else:\n",
    "        voice = \"-\"\n",
    "    \n",
    "    masculine = re.search(\"gen1\", ITtag)\n",
    "    feminine = re.search(\"gen2\", ITtag)\n",
    "    neuter = re.search(\"gen3\", ITtag)\n",
    "    if masculine:\n",
    "        gender = \"m\"\n",
    "    elif feminine:\n",
    "        gender = \"f\"\n",
    "    elif neuter:\n",
    "        gender = \"n\"\n",
    "    else:\n",
    "        gender = \"-\"\n",
    "        \n",
    "    nominative = re.search(\"casA|casJ\", ITtag)\n",
    "    genitive = re.search(\"casB|casK\", ITtag)\n",
    "    dative = re.search(\"casC|casL\", ITtag)\n",
    "    accusative = re.search(\"casD|casM\", ITtag)\n",
    "    vocative = re.search(\"casE|casN\", ITtag)\n",
    "    ablative = re.search(\"casF|casO\", ITtag)\n",
    "    if nominative:\n",
    "        case = \"n\"\n",
    "    elif genitive:\n",
    "        case = \"g\"\n",
    "    elif dative:\n",
    "        case = \"d\"\n",
    "    elif accusative:\n",
    "        case = \"a\"\n",
    "    elif vocative:\n",
    "        case = \"v\"\n",
    "    elif ablative:\n",
    "        case = \"b\"\n",
    "    else:\n",
    "        case = \"-\"\n",
    "    \n",
    "    positive = re.search(\"'Degree': 'Pos'\", ITtag) \n",
    "    comparative = re.search(\"'Degree': 'Cmp'\", ITtag)\n",
    "    superlative = re.search(\"'Degree': 'Sup'\", ITtag)\n",
    "    if positive and (pos == \"a\"):\n",
    "        degree = \"p\"\n",
    "    elif comparative:\n",
    "        degree = \"c\"\n",
    "    elif superlative:\n",
    "        degree = \"s\"\n",
    "    else:\n",
    "        degree = \"-\"\n",
    "\n",
    "    IT_new_tag = f\"{pos}.{person}.{number}.{tense}.{mood}.{voice}.{gender}.{case}.{degree}\"\n",
    "    return IT_new_tag"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As Universal Dependencies does not distinguish between passive and deponent verbs, tokens are assigned the \"deponent\" tag by matching the lemma with the following list of deponent verbs, taken from en.wiktionary.org/wiki/Category:Latin_deponent_verbs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "deponent_list_jv = \"abominor, aborior, aboriscor, abutor, acedior, adfector, adfor, adgeniculor, adgredior, adhortor, adipiscor, adloquor, adluctor, admetior, admiror, admoderor, admodulor, admurmuror, adnitor, adopinor, adorior, adosculor, adprecor, adsector, adsellor, adsentior, adsentor, adsequor, adstipulor, adtestor, adulescentior, adulor, adutor, adveneror, adversor, adzelor, aeditumor, aemulor, affector, affor, aggeniculor, aggredior, agnascor, agricolor, allabor, alloquor, alluctor, altercor, alucinor, alumnor, amolior, amplector, amplexor, ampullor, ancillor, annitor, annonor, antegredior, antestor, apiscor, aporior, apprecor, apricor, aquor, arbitror, architector, argumentor, argutor, ariolor, aspernor, assector, assellor, assentior, assentor, assequor, astipulor, attestor, auctionor, aucupor, auspicor, auxilior, aversor, bacchor, baccor, baubor, belligeror, bellor, blandior, bovinor, calumnior, calvor, castrametor, causor, cavillor, certor, circumgredior, circumloquor, circumtueor, circumvehor, clueor, cohortor, collabor, colloquor, colluctor, comissor, comitor, commentor, commereor, commetior, comminiscor, comminor, commiseror, commorior, commoror, compaciscor, compartior, compatior, compeciscor, complector, compluor, concionor, concomitor, condelector, confabulor, confiteor, congratulor, congredior, conitor, coniucundor, conlabor, conloquor, conluctor, conmetior, conmorior, connitor, conor, conpartior, conpatior, conplector, conpluor, conqueror, conscreor, consector, consequor, consermonor, consilior, consolor, conspicor, contemplor, contestor, contionor, contueor, contutor, convescor, convicior, convivor, cooperor, coorior, coutor, crapulor, criminor, cunctor, debacchor, defungor, degredior, dehortor, delabor, demereor, demetior, demiror, demolior, demorior, demoror, deosculor, depeciscor, depeculor, depopulor, depraedor, deprecor, detestor, deutor, deversor, diffiteor, digredior, dilabor, dilargior, dimetior, divagor, dominor, eblandior, effor, egredior, eiaculor, elabor, eloquor, eluctor, ementior, emercor, emereor, emetior, emiror, emodulor, emolior, emorior, enascor, enitor, epulor, evagor, examplexor, execror, exequor, exhortor, exordior, exorior, expergiscor, experior, expiscor, exsecror, exsequor, fabulor, famulor, fateor, fatuor, ferior, for, fornicor, frumentor, fruor, fungor, furor, galaticor, glorior, gnascor, gradior, grassor, gratificor, grator, gratulor, halucinor, hariolor, heiulor, hejulor, helluor, horior, hortor, hospitor, iaculor, illabor, imaginor, imitor, immoror, imprecor, indignor, infitior, ingredior, inhortor, iniurior, injurior, innitor, inoblector, inoperor, inprecor, insector, insequor, insidior, interfor, interloquor, interminor, intermorior, interpretor, introgredior, intueor, iocor, ioculor, irascor, itineror, iucundor, jaculor, joculor, jucundor, labor, laetor, lamentor, largior, latinor, latrocinor, liceor, lignor, liquor, locor, loquor, lucror, luctor, lupor, machinor, manuor, masturbor, materior, medeor, meditor, memoror, mentior, mercor, mereor, metior, metor, minitor, minor, miror, misereor, miseror, moderor, modulor, moechor, molior, morior, moror, mutuor, nanciscor, nascor, natinor, negotior, nepotor, nidulor, nitor, nixor, nugor, nundinor, obaemulor, obliviscor, obloquor, obnitor, oborior, obsequor, obtestor, obtueor, obversor, ominor, operor, opinor, opitulor, opperior, ordior, orior, oscitor, osculor, otior, pabulor, paciscor, palor, pandiculor, partior, pascor, patior, patrocinor, peculor, perbacchor, perconor, percontor, peregrinor, perfruor, perfungor, periclitor, perlabor, permetior, perpetior, perscrutor, persector, persequor, perspeculor, pervagor, philosophor, pigritor, piscor, polliceor, pollicitor, populor, potior, praedominor, praedor, praefor, praegredior, praeloquor, praemeditor, praemorior, praestolor, praetergredior, praeterlabor, praetervehor, praevaricor, praevehor, precor, principor, procor, proelior, proficiscor, profiteor, profor, progredior, prolabor, proloquor, promereor, promutuor, prosequor, protestor, pugillor, pugilor, queritor, queror, quietor, racemor, ratiocinor, recordor, refragor, regredior, relabor, reliquor, reluctor, rememoror, remetior, reminiscor, remoror, renascor, renitor, reor, resequor, restipulor, retrogradior, revereor, revertor, rimor, ringor, rixor, ruminor, ruspor, rusticor, savior, sciscitor, scitor, scortor, scrutor, sector, sequor, sermocinor, sermonor, simplificor, solor, sortior, spatior, speculor, stipulor, stomachor, suavior, subaemulor, subpetior, subsequor, subsortior, subvereor, suffuror, supergredior, superlucror, supervehor, suppetior, suspicor, taurobolior, tergiversor, testificor, testor, traloquor, transgredior, tricor, tristor, tueor, tumultuor, tutor, ulciscor, urinor, usitor, utor, vador, vagor, vaticinor, velificor, velitor, veneror, venor, vereor, vermiculor, versor, vescor, vicinor, vociferor, vulpinor\"\n",
    "deponent_list_iv = deponent_list_jv.replace(\"j\", \"i\")\n",
    "deponent_list_iu = deponent_list_iv.replace(\"v\", \"u\")\n",
    "deponent_list_jv = deponent_list_jv.split(\", \")\n",
    "deponent_list_iv = deponent_list_iv.split(\", \")\n",
    "deponent_list_iu = deponent_list_iu.split(\", \")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Preparing the LLCT training data. The CoNLL-U file is read with the conllu package (github.com/EmilStenstrom/conllu), then the tokens and lemmata are extracted, and the tags converted with the function defined above. These three lists are then combined as a Pandas dataframe and saved as a tab seperated file (required by the RFTagger)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from conllu import parse\n",
    "LLCT_dev = parse(open(\"la_llct-ud-dev.conllu\").read())\n",
    "LLCT_train = parse(open(\"la_llct-ud-train.conllu\").read())\n",
    "\n",
    "LLCT_devtokens = []\n",
    "LLCT_devlemmata = []\n",
    "LLCT_devtags = []\n",
    "LLCT_traintokens = []\n",
    "LLCT_trainlemmata = []\n",
    "LLCT_traintags = []\n",
    "\n",
    "for i in range(len(LLCT_dev)):\n",
    "    sentence = LLCT_dev[i]\n",
    "    for j in range(len(sentence)):\n",
    "        token = sentence[j]\n",
    "        form = token[\"form\"]\n",
    "        lemma = token[\"lemma\"]\n",
    "        upos = token[\"upos\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        feats = token[\"feats\"]\n",
    "        LLCTdata = f\"{lemma} {upos} {xpos} {feats}\"\n",
    "        LLCT_devtokens.append(form)\n",
    "        LLCT_devlemmata.append(lemma)\n",
    "        LLCT_devtags.append(LLCTdata)\n",
    "\n",
    "for k in range(len(LLCT_train)):\n",
    "    sentence = LLCT_train[k]\n",
    "    for l in range(len(sentence)):\n",
    "        token = sentence[l]\n",
    "        form = token[\"form\"]\n",
    "        lemma = token[\"lemma\"]\n",
    "        upos = token[\"upos\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        feats = token[\"feats\"]\n",
    "        LLCTdata = f\"{lemma} {upos} {xpos} {feats}\"\n",
    "        LLCT_traintokens.append(form)\n",
    "        LLCT_trainlemmata.append(lemma)\n",
    "        LLCT_traintags.append(LLCTdata)\n",
    "\n",
    "LLCTtokens = LLCT_devtokens + LLCT_traintokens\n",
    "LLCTlemmata = LLCT_devlemmata + LLCT_trainlemmata\n",
    "LLCTtags = LLCT_devtags + LLCT_traintags\n",
    "\n",
    "LLCT_new_tags = []\n",
    "for m in range(len(LLCTtags)):\n",
    "    LLCTtag = LLCTtags[m]\n",
    "    LLCTlemma = LLCTlemmata[m]\n",
    "    LLCT_new_tag = LLCTtoPerseus(LLCTtag, LLCTlemma)\n",
    "    LLCT_new_tags.append(LLCT_new_tag)\n",
    "\n",
    "import pandas as pd\n",
    "LLCT_df = pd.DataFrame()\n",
    "LLCT_df[\"Token\"] = LLCTtokens\n",
    "LLCT_df[\"Tag\"] = LLCT_new_tags\n",
    "LLCT_df[\"Lemma\"] = LLCTlemmata\n",
    "LLCT_df.to_csv(\"LLCT-training-corpus.txt\",sep=\"\\t\", header = False, index = False)\n",
    "LLCT_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Preparing the PROIEL training data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from io import open\n",
    "from conllu import parse_incr\n",
    "\n",
    "PROIEL_devtokens = []\n",
    "PROIEL_devlemmata = []\n",
    "PROIEL_devtags = []\n",
    "PROIEL_traintokens = []\n",
    "PROIEL_trainlemmata = []\n",
    "PROIEL_traintags = []\n",
    "PROIEL_testtokens = []\n",
    "PROIEL_testlemmata = []\n",
    "PROIEL_testtags = []\n",
    "\n",
    "data_file = open(\"la_proiel-ud-dev.conllu\", \"r\", encoding=\"utf-8\")\n",
    "for tokenlist in parse_incr(data_file):\n",
    "    sentence = tokenlist\n",
    "    sentence.append({\"form\": \".\", \"lemma\": \".\",\"xpos\": \"PUNCT\"})\n",
    "    for i in range(len(sentence)):\n",
    "        token = sentence[i]\n",
    "        form = token[\"form\"]\n",
    "        lemma = token[\"lemma\"]\n",
    "        upos = token[\"upos\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        feats = token[\"feats\"]\n",
    "        PROIELdata = f\"{lemma} {upos} {xpos} {feats}\"\n",
    "        PROIEL_devtokens.append(form)\n",
    "        PROIEL_devlemmata.append(lemma)\n",
    "        PROIEL_devtags.append(PROIELdata)\n",
    "        \n",
    "data_file2 = open(\"la_proiel-ud-train.conllu\", \"r\", encoding=\"utf-8\")\n",
    "for tokenlist in parse_incr(data_file2):\n",
    "    sentence = tokenlist\n",
    "    sentence.append({\"form\": \".\", \"lemma\": \".\",\"xpos\": \"PUNCT\"})\n",
    "    for j in range(len(sentence)):\n",
    "        token = sentence[j]\n",
    "        form = token[\"form\"]\n",
    "        lemma = token[\"lemma\"]\n",
    "        upos = token[\"upos\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        feats = token[\"feats\"]\n",
    "        PROIELdata = f\"{lemma} {upos} {xpos} {feats}\"\n",
    "        PROIEL_traintokens.append(form)\n",
    "        PROIEL_trainlemmata.append(lemma)\n",
    "        PROIEL_traintags.append(PROIELdata)\n",
    "\n",
    "data_file3 = open(\"la_proiel-ud-test.conllu\", \"r\", encoding=\"utf-8\")\n",
    "for tokenlist in parse_incr(data_file3):\n",
    "    sentence = tokenlist\n",
    "    sentence.append({\"form\": \".\", \"lemma\": \".\",\"xpos\": \"PUNCT\"})\n",
    "    for k in range(len(sentence)):\n",
    "        token = sentence[k]\n",
    "        form = token[\"form\"]\n",
    "        lemma = token[\"lemma\"]\n",
    "        upos = token[\"upos\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        feats = token[\"feats\"]\n",
    "        PROIELdata = f\"{lemma} {upos} {xpos} {feats}\"\n",
    "        PROIEL_testtokens.append(form)\n",
    "        PROIEL_testlemmata.append(lemma)\n",
    "        PROIEL_testtags.append(PROIELdata)\n",
    "PROIELtokens = PROIEL_devtokens + PROIEL_traintokens + PROIEL_testtokens\n",
    "PROIELlemmata = PROIEL_devlemmata + PROIEL_trainlemmata + PROIEL_testlemmata\n",
    "PROIELtags = PROIEL_devtags + PROIEL_traintags + PROIEL_testtags\n",
    "\n",
    "PROIEL_new_tags = []\n",
    "for m in range(len(PROIELtags)):\n",
    "    PROIELtag = PROIELtags[m]\n",
    "    PROIELlemma = PROIELlemmata[m]\n",
    "    PROIEL_new_tag = PROIELtoPerseus(PROIELtag, PROIELlemma)\n",
    "    PROIEL_new_tags.append(PROIEL_new_tag)\n",
    "import pandas as pd\n",
    "PROIEL_df = pd.DataFrame()\n",
    "PROIEL_df[\"Token\"] = PROIELtokens\n",
    "PROIEL_df[\"Tag\"] = PROIEL_new_tags\n",
    "PROIEL_df[\"Lemma\"] = PROIELlemmata\n",
    "PROIEL_df.to_csv(\"PROIEL-train-corpus.txt\",sep=\"\\t\", header = False, index = False)\n",
    "PROIEL_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The Perseus Latin Dependency Treebank requires no tagset conversion."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from conllu import parse\n",
    "Perseus_test = parse(open(\"la_perseus-ud-test.conllu\").read())\n",
    "Perseus_train = parse(open(\"la_perseus-ud-train.conllu\").read())\n",
    "\n",
    "Perseus_traintokens = []\n",
    "Perseus_trainlemmata = []\n",
    "Perseus_traintags = []\n",
    "Perseus_testtokens = []\n",
    "Perseus_testlemmata = []\n",
    "Perseus_testtags = []\n",
    "\n",
    "for i in range(len(Perseus_train)):\n",
    "    sentence = Perseus_train[i]\n",
    "    for j in range(len(sentence)):\n",
    "        token = sentence[j]\n",
    "        lemma = token[\"lemma\"]\n",
    "        form = token[\"form\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        a = xpos[0]\n",
    "        b = xpos[1]\n",
    "        c = xpos[2]\n",
    "        d = xpos[3]\n",
    "        e = xpos[4]\n",
    "        f = xpos[5]\n",
    "        g = xpos[6]\n",
    "        h = xpos[7]\n",
    "        i = xpos[8]\n",
    "        tag = f\"{a}.{b}.{c}.{d}.{e}.{f}.{g}.{h}.{i}\"\n",
    "        Perseus_traintokens.append(form)\n",
    "        Perseus_trainlemmata.append(lemma)\n",
    "        Perseus_traintags.append(tag)\n",
    "\n",
    "for k in range(len(Perseus_test)):\n",
    "    sentence = Perseus_test[k]\n",
    "    for l in range(len(sentence)):\n",
    "        token = sentence[l]\n",
    "        lemma = token[\"lemma\"]\n",
    "        form = token[\"form\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        a = xpos[0]\n",
    "        b = xpos[1]\n",
    "        c = xpos[2]\n",
    "        d = xpos[3]\n",
    "        e = xpos[4]\n",
    "        f = xpos[5]\n",
    "        g = xpos[6]\n",
    "        h = xpos[7]\n",
    "        i = xpos[8]\n",
    "        tag = f\"{a}.{b}.{c}.{d}.{e}.{f}.{g}.{h}.{i}\"\n",
    "        Perseus_testtokens.append(form)\n",
    "        Perseus_testlemmata.append(lemma)\n",
    "        Perseus_testtags.append(tag)\n",
    "\n",
    "Perseus_tokens = Perseus_testtokens + Perseus_traintokens\n",
    "Perseus_lemmata = Perseus_testlemmata + Perseus_trainlemmata\n",
    "Perseus_tags = Perseus_testtags + Perseus_traintags\n",
    "\n",
    "Perseus_df = pd.DataFrame()\n",
    "Perseus_df[\"Token\"] = Perseus_tokens\n",
    "Perseus_df[\"Tag\"] = Perseus_tags\n",
    "Perseus_df[\"Lemma\"] = Perseus_lemmata\n",
    "Perseus_df.to_csv(\"Perseus-training-corpus.txt\",sep=\"\\t\", header = False, index = False)\n",
    "Perseus_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Preparing the _Index Thomisiticus_ training fold."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from conllu import parse\n",
    "IT_dev = parse(open(\"la_ittb-ud-dev.conllu\").read())\n",
    "IT_train = parse(open(\"la_ittb-ud-train.conllu\").read())\n",
    "\n",
    "IT_devtokens = []\n",
    "IT_devlemmata = []\n",
    "IT_devtags = []\n",
    "IT_traintokens = []\n",
    "IT_trainlemmata = []\n",
    "IT_traintags = []\n",
    "\n",
    "for i in range(len(IT_dev)):\n",
    "    sentence = IT_dev[i]\n",
    "    for j in range(len(sentence)):\n",
    "        token = sentence[j]\n",
    "        form = token[\"form\"]\n",
    "        lemma = token[\"lemma\"]\n",
    "        upos = token[\"upos\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        feats = token[\"feats\"]\n",
    "        ITdata = f\"{lemma} {upos} {xpos} {feats}\"\n",
    "        IT_devtokens.append(form)\n",
    "        IT_devlemmata.append(lemma)\n",
    "        IT_devtags.append(ITdata)\n",
    "\n",
    "for k in range(len(IT_train)):\n",
    "    sentence = IT_train[k]\n",
    "    for l in range(len(sentence)):\n",
    "        token = sentence[l]\n",
    "        form = token[\"form\"]\n",
    "        lemma = token[\"lemma\"]\n",
    "        upos = token[\"upos\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        feats = token[\"feats\"]\n",
    "        ITdata = f\"{lemma} {upos} {xpos} {feats}\"\n",
    "        IT_traintokens.append(form)\n",
    "        IT_trainlemmata.append(lemma)\n",
    "        IT_traintags.append(ITdata)\n",
    "\n",
    "ITtokens = IT_devtokens + IT_traintokens\n",
    "ITlemmata = IT_devlemmata + IT_trainlemmata\n",
    "ITtags = IT_devtags + IT_traintags\n",
    "\n",
    "IT_new_tags = []\n",
    "for m in range(len(ITtags)):\n",
    "    ITtag = ITtags[m]\n",
    "    ITlemma = ITlemmata[m]\n",
    "    IT_new_tag = ITtoPerseus(ITtag, ITlemma)\n",
    "    IT_new_tags.append(IT_new_tag)\n",
    "import pandas as pd\n",
    "IT_df = pd.DataFrame()\n",
    "IT_df[\"Token\"] = ITtokens\n",
    "IT_df[\"Tag\"] = IT_new_tags\n",
    "IT_df[\"Lemma\"] = ITlemmata\n",
    "IT_df.to_csv(\"IT-training-corpus.txt\",sep=\"\\t\", header = False, index = False)\n",
    "IT_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Preparing the testing fold. A similar process was carried out on the GCSE data sets, but they cannot be shown here."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "IT_test = parse(open(\"la_ittb-ud-test.conllu\").read())\n",
    "IT_testtokens = []\n",
    "IT_testlemmata = []\n",
    "IT_testtags = []\n",
    "\n",
    "for n in range(len(IT_test)):\n",
    "    sentence = IT_test[n]\n",
    "    for o in range(len(sentence)):\n",
    "        token = sentence[o]\n",
    "        form = token[\"form\"]\n",
    "        lemma = token[\"lemma\"]\n",
    "        upos = token[\"upos\"]\n",
    "        xpos = token[\"xpos\"]\n",
    "        feats = token[\"feats\"]\n",
    "        ITdata = f\"{lemma} {upos} {xpos} {feats}\"\n",
    "        IT_testtokens.append(form)\n",
    "        IT_testlemmata.append(lemma)\n",
    "        IT_testtags.append(ITdata)\n",
    "IT_test_new_tags = []\n",
    "\n",
    "for s in range(len(IT_testtags)):\n",
    "    ITtag = IT_testtags[s]\n",
    "    ITlemma = IT_testlemmata[s]\n",
    "    IT_new_tag = ITtoPerseus(ITtag, ITlemma)\n",
    "    IT_test_new_tags.append(IT_new_tag)\n",
    "IT_df2 = pd.DataFrame()\n",
    "IT_df2[\"Token\"] = IT_testtokens\n",
    "IT_df2[\"Tag\"] = IT_test_new_tags\n",
    "IT_df2[\"Lemma\"] = IT_testlemmata\n",
    "IT_df2.to_csv(\"IT-test-corpus.txt\",sep=\"\\t\", header = False, index = False)\n",
    "IT_df2.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For each test set, there is a ground truth file with token, manual tag and lemma. Here the tag and lemma are removed. Line breaks were added in afterwards. Carriage returns also need to be removed for the RFTagger to accept a file."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "IT_gt =  pd.read_csv(\"IT-test-gt.txt\", sep = \"\\t\", header = None, index_col = False, na_filter = False)\n",
    "IT_test = IT_gt[0]\n",
    "IT_test.to_csv(\"IT-test.txt\", header = False, index = False, line_terminator = \"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generating models and predictions with the RFTagger\n",
    "\n",
    "The RFTagger runs in Linux, and so cannot be shown here, but the process for operating it is as follows:\n",
    "\n",
    "#### Installation\n",
    "\n",
    "I installed the tagger from Winge's github page and followed the instructions there (github.com/Alatius/latin-macronizer).\n",
    "\n",
    "#### Training \n",
    "\n",
    "The command to train a model is `rft-train -l [lexicon .txt file] -c [value] [training data as .txt file] wordclass.txt [model name .model file` Thus `rft-train -l rftagger-lexicon.txt -c 2 LLCT-training-corpus.txt wordclass.txt LL02.model` generated the LL02 model (`LL02.model`) using the LLCT training corpus, the external lexicon and a unigram context window.\n",
    "\n",
    "#### Annotating\n",
    "\n",
    "To annotate a new text with model, the text must be already tokenised with one token per line and an empty line after each sentence. The command then is `rft-annotate -s [model name .model file] [new text as .txt file] [output text as .txt file]` For example: `rft-annotate -s Max10.model IT-test-corpus.txt IT-predict-Max10.txt` The `-s` option has the tagger look up sentence intial words in both lower-case and upper-case form. Necessary for Latin as capitalisation norms are inconsistent across publishers and periods."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Calculating predictive performance and plotting results \n",
    "\n",
    "The predictions are loaded into a dataframe and then tested against the ground truth tag."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Token</th>\n",
       "      <th>GT Tag</th>\n",
       "      <th>Lemma</th>\n",
       "      <th>CL02</th>\n",
       "      <th>CL03</th>\n",
       "      <th>CL04</th>\n",
       "      <th>CL05</th>\n",
       "      <th>CL06</th>\n",
       "      <th>CL07</th>\n",
       "      <th>CL08</th>\n",
       "      <th>...</th>\n",
       "      <th>ML10</th>\n",
       "      <th>Max02</th>\n",
       "      <th>Max03</th>\n",
       "      <th>Max04</th>\n",
       "      <th>Max05</th>\n",
       "      <th>Max06</th>\n",
       "      <th>Max07</th>\n",
       "      <th>Max08</th>\n",
       "      <th>Max09</th>\n",
       "      <th>Max10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>stabat</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>sto</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>...</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "      <td>v.3.s.i.i.a.-.-.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>pro</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>pro</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>...</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>litore</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>litus</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>...</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>diversa</td>\n",
       "      <td>a.-.s.-.-.-.f.n.p</td>\n",
       "      <td>diversus</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>...</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.b.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.b.-</td>\n",
       "      <td>v.-.s.r.p.p.f.b.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "      <td>v.-.s.r.p.p.f.n.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>acies</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>acies</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>...</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.s.-.-.-.f.n.-</td>\n",
       "      <td>n.-.p.-.-.-.f.a.-</td>\n",
       "      <td>n.-.p.-.-.-.f.a.-</td>\n",
       "      <td>n.-.p.-.-.-.f.a.-</td>\n",
       "      <td>n.-.p.-.-.-.f.a.-</td>\n",
       "      <td>n.-.p.-.-.-.f.a.-</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 39 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     Token             GT Tag     Lemma               CL02               CL03  \\\n",
       "0   stabat  v.3.s.i.i.a.-.-.-       sto  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-   \n",
       "1      pro  r.-.-.-.-.-.-.-.-       pro  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "2   litore  n.-.s.-.-.-.n.b.-     litus  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-   \n",
       "3  diversa  a.-.s.-.-.-.f.n.p  diversus  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.n.-   \n",
       "4    acies  n.-.s.-.-.-.f.n.-     acies  n.-.s.-.-.-.f.n.-  n.-.s.-.-.-.f.n.-   \n",
       "\n",
       "                CL04               CL05               CL06               CL07  \\\n",
       "0  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-   \n",
       "1  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "2  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-   \n",
       "3  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.n.-   \n",
       "4  n.-.s.-.-.-.f.n.-  n.-.s.-.-.-.f.n.-  n.-.s.-.-.-.f.n.-  n.-.s.-.-.-.f.n.-   \n",
       "\n",
       "                CL08  ...               ML10              Max02  \\\n",
       "0  v.3.s.i.i.a.-.-.-  ...  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-   \n",
       "1  r.-.-.-.-.-.-.-.-  ...  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "2  n.-.s.-.-.-.n.b.-  ...  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-   \n",
       "3  v.-.s.r.p.p.f.n.-  ...  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.b.-   \n",
       "4  n.-.s.-.-.-.f.n.-  ...  n.-.s.-.-.-.f.n.-  n.-.s.-.-.-.f.n.-   \n",
       "\n",
       "               Max03              Max04              Max05              Max06  \\\n",
       "0  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-   \n",
       "1  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "2  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-   \n",
       "3  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.b.-   \n",
       "4  n.-.s.-.-.-.f.n.-  n.-.s.-.-.-.f.n.-  n.-.s.-.-.-.f.n.-  n.-.p.-.-.-.f.a.-   \n",
       "\n",
       "               Max07              Max08              Max09              Max10  \n",
       "0  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-  v.3.s.i.i.a.-.-.-  \n",
       "1  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  \n",
       "2  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-  \n",
       "3  v.-.s.r.p.p.f.b.-  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.n.-  v.-.s.r.p.p.f.n.-  \n",
       "4  n.-.p.-.-.-.f.a.-  n.-.p.-.-.-.f.a.-  n.-.p.-.-.-.f.a.-  n.-.p.-.-.-.f.a.-  \n",
       "\n",
       "[5 rows x 39 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "Boudica_prediction_names = [\"Boudica-predict-CL02.txt\", \"Boudica-predict-CL03.txt\", \"Boudica-predict-CL04.txt\", \"Boudica-predict-CL05.txt\", \"Boudica-predict-CL06.txt\", \"Boudica-predict-CL07.txt\", \"Boudica-predict-CL08.txt\", \"Boudica-predict-CL09.txt\", \"Boudica-predict-CL10.txt\", \"Boudica-predict-LL02.txt\", \"Boudica-predict-LL03.txt\", \"Boudica-predict-LL04.txt\", \"Boudica-predict-LL05.txt\", \"Boudica-predict-LL06.txt\", \"Boudica-predict-LL07.txt\", \"Boudica-predict-LL08.txt\", \"Boudica-predict-LL09.txt\", \"Boudica-predict-LL10.txt\", \"Boudica-predict-ML02.txt\", \"Boudica-predict-ML03.txt\", \"Boudica-predict-ML04.txt\", \"Boudica-predict-ML05.txt\", \"Boudica-predict-ML06.txt\", \"Boudica-predict-ML07.txt\", \"Boudica-predict-ML08.txt\", \"Boudica-predict-ML09.txt\", \"Boudica-predict-ML10.txt\", \"Boudica-predict-Max02.txt\", \"Boudica-predict-Max03.txt\", \"Boudica-predict-Max04.txt\", \"Boudica-predict-Max05.txt\", \"Boudica-predict-Max06.txt\", \"Boudica-predict-Max07.txt\", \"Boudica-predict-Max08.txt\", \"Boudica-predict-Max09.txt\", \"Boudica-predict-Max10.txt\"]\n",
    "\n",
    "column_names = [\"Token\", \"GT Tag\", \"Lemma\", \"CL02\", \"CL03\", \"CL04\", \"CL05\", \"CL06\", \"CL07\", \"CL08\", \"CL09\", \"CL10\", \"LL02\", \"LL03\", \"LL04\", \"LL05\", \"LL06\", \"LL07\", \"LL08\", \"LL09\", \"LL10\", \"ML02\", \"ML03\", \"ML04\", \"ML05\", \"ML06\", \"ML07\", \"ML08\", \"ML09\", \"ML10\", \"Max02\", \"Max03\", \"Max04\", \"Max05\", \"Max06\", \"Max07\", \"Max08\", \"Max09\", \"Max10\"]\n",
    "df1 = pd.read_csv(\"Boudica-gt.txt\", sep = \"\\t\", header = None, index_col = False, na_filter = False)\n",
    "\n",
    "for i in Boudica_prediction_names:\n",
    "    df_i = pd.read_csv(f\"{i}\", sep = \"\\t\", header = None, index_col = False, na_filter = False)\n",
    "    df_i = df_i[1]\n",
    "    df1 = pd.concat([df1, df_i], axis = 1)\n",
    "df1.columns = column_names\n",
    "df1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CL02</th>\n",
       "      <th>CL03</th>\n",
       "      <th>CL04</th>\n",
       "      <th>CL05</th>\n",
       "      <th>CL06</th>\n",
       "      <th>CL07</th>\n",
       "      <th>CL08</th>\n",
       "      <th>CL09</th>\n",
       "      <th>CL10</th>\n",
       "      <th>LL02</th>\n",
       "      <th>...</th>\n",
       "      <th>ML10</th>\n",
       "      <th>Max02</th>\n",
       "      <th>Max03</th>\n",
       "      <th>Max04</th>\n",
       "      <th>Max05</th>\n",
       "      <th>Max06</th>\n",
       "      <th>Max07</th>\n",
       "      <th>Max08</th>\n",
       "      <th>Max09</th>\n",
       "      <th>Max10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 36 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    CL02   CL03   CL04   CL05   CL06   CL07   CL08   CL09   CL10   LL02  ...  \\\n",
       "0   True   True   True   True   True   True   True   True   True   True  ...   \n",
       "1   True   True   True   True   True   True   True   True   True   True  ...   \n",
       "2   True   True   True   True   True   True   True   True   True   True  ...   \n",
       "3  False  False  False  False  False  False  False  False  False  False  ...   \n",
       "4   True   True   True   True   True   True   True   True   True  False  ...   \n",
       "\n",
       "    ML10  Max02  Max03  Max04  Max05  Max06  Max07  Max08  Max09  Max10  \n",
       "0   True   True   True   True   True   True   True   True   True   True  \n",
       "1   True   True   True   True   True   True   True   True   True   True  \n",
       "2   True   True   True   True   True   True   True   True   True   True  \n",
       "3  False  False  False  False  False  False  False  False  False  False  \n",
       "4   True   True   True   True   True  False  False  False  False  False  \n",
       "\n",
       "[5 rows x 36 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "df2 = pd.DataFrame()\n",
    "\n",
    "\n",
    "for column in df1.columns[3:]:\n",
    "    df2[column] = np.where(df1[column]== df1[\"GT Tag\"], True, False)\n",
    "df2.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CL02     66.6\n",
      "CL03     67.2\n",
      "CL04     67.2\n",
      "CL05     67.3\n",
      "CL06     67.3\n",
      "CL07     67.0\n",
      "CL08     67.2\n",
      "CL09     66.8\n",
      "CL10     67.2\n",
      "LL02     56.5\n",
      "LL03     57.9\n",
      "LL04     57.6\n",
      "LL05     56.9\n",
      "LL06     59.6\n",
      "LL07     59.1\n",
      "LL08     59.2\n",
      "LL09     59.1\n",
      "LL10     58.9\n",
      "ML02     63.5\n",
      "ML03     63.3\n",
      "ML04     64.2\n",
      "ML05     64.6\n",
      "ML06     64.1\n",
      "ML07     63.8\n",
      "ML08     63.3\n",
      "ML09     63.3\n",
      "ML10     63.5\n",
      "Max02    65.9\n",
      "Max03    66.5\n",
      "Max04    65.8\n",
      "Max05    65.0\n",
      "Max06    65.8\n",
      "Max07    66.5\n",
      "Max08    66.3\n",
      "Max09    66.5\n",
      "Max10    66.2\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "Boudica_acc = df2.sum(axis = 0)/df2.shape[0]\n",
    "Boudica_acc = Boudica_acc.multiply(100).round(decimals = 1)\n",
    "print(Boudica_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Token</th>\n",
       "      <th>GT Tag</th>\n",
       "      <th>Lemma</th>\n",
       "      <th>CL02</th>\n",
       "      <th>CL03</th>\n",
       "      <th>CL04</th>\n",
       "      <th>CL05</th>\n",
       "      <th>CL06</th>\n",
       "      <th>CL07</th>\n",
       "      <th>CL08</th>\n",
       "      <th>...</th>\n",
       "      <th>ML10</th>\n",
       "      <th>Max02</th>\n",
       "      <th>Max03</th>\n",
       "      <th>Max04</th>\n",
       "      <th>Max05</th>\n",
       "      <th>Max06</th>\n",
       "      <th>Max07</th>\n",
       "      <th>Max08</th>\n",
       "      <th>Max09</th>\n",
       "      <th>Max10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>consilii</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>consilium</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>...</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "      <td>n.-.s.-.-.-.n.g.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>tamen</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>tamen</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>...</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ante</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>ante</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>...</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>sui</td>\n",
       "      <td>a.-.s.-.-.-.n.g.p</td>\n",
       "      <td>suus</td>\n",
       "      <td>a.-.p.-.-.-.n.n.-</td>\n",
       "      <td>p.3.s.-.-.-.m.g.-</td>\n",
       "      <td>p.3.s.-.-.-.m.g.-</td>\n",
       "      <td>a.-.p.-.-.-.n.n.-</td>\n",
       "      <td>a.-.p.-.-.-.n.n.-</td>\n",
       "      <td>a.-.p.-.-.-.n.n.-</td>\n",
       "      <td>a.-.p.-.-.-.n.n.-</td>\n",
       "      <td>...</td>\n",
       "      <td>a.-.s.-.-.-.n.g.p</td>\n",
       "      <td>a.-.p.-.-.-.m.b.-</td>\n",
       "      <td>a.-.s.-.-.-.f.b.p</td>\n",
       "      <td>a.-.s.-.-.-.f.b.p</td>\n",
       "      <td>a.-.s.-.-.-.f.b.p</td>\n",
       "      <td>a.-.s.-.-.-.f.b.p</td>\n",
       "      <td>a.-.s.-.-.-.f.b.p</td>\n",
       "      <td>a.-.s.-.-.-.f.b.p</td>\n",
       "      <td>a.-.s.-.-.-.f.b.p</td>\n",
       "      <td>a.-.s.-.-.-.f.b.p</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>fidissima</td>\n",
       "      <td>a.-.s.-.-.-.f.v.p</td>\n",
       "      <td>fidissimus</td>\n",
       "      <td>a.-.p.-.-.-.n.n.s</td>\n",
       "      <td>a.-.p.-.-.-.n.a.s</td>\n",
       "      <td>a.-.p.-.-.-.n.a.s</td>\n",
       "      <td>a.-.p.-.-.-.n.n.s</td>\n",
       "      <td>a.-.p.-.-.-.n.n.s</td>\n",
       "      <td>a.-.p.-.-.-.n.n.s</td>\n",
       "      <td>a.-.p.-.-.-.n.n.s</td>\n",
       "      <td>...</td>\n",
       "      <td>a.-.p.-.-.-.n.a.s</td>\n",
       "      <td>a.-.p.-.-.-.n.a.s</td>\n",
       "      <td>a.-.s.-.-.-.f.n.s</td>\n",
       "      <td>a.-.s.-.-.-.f.n.s</td>\n",
       "      <td>a.-.s.-.-.-.f.n.s</td>\n",
       "      <td>a.-.s.-.-.-.f.n.s</td>\n",
       "      <td>a.-.s.-.-.-.f.n.s</td>\n",
       "      <td>a.-.s.-.-.-.f.n.s</td>\n",
       "      <td>a.-.s.-.-.-.f.n.s</td>\n",
       "      <td>a.-.s.-.-.-.f.n.s</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 39 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Token             GT Tag       Lemma               CL02  \\\n",
       "0   consilii  n.-.s.-.-.-.n.g.-   consilium  n.-.s.-.-.-.n.g.-   \n",
       "1      tamen  d.-.-.-.-.-.-.-.-       tamen  d.-.-.-.-.-.-.-.-   \n",
       "2       ante  d.-.-.-.-.-.-.-.-        ante  d.-.-.-.-.-.-.-.-   \n",
       "3        sui  a.-.s.-.-.-.n.g.p        suus  a.-.p.-.-.-.n.n.-   \n",
       "4  fidissima  a.-.s.-.-.-.f.v.p  fidissimus  a.-.p.-.-.-.n.n.s   \n",
       "\n",
       "                CL03               CL04               CL05               CL06  \\\n",
       "0  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-   \n",
       "1  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "2  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "3  p.3.s.-.-.-.m.g.-  p.3.s.-.-.-.m.g.-  a.-.p.-.-.-.n.n.-  a.-.p.-.-.-.n.n.-   \n",
       "4  a.-.p.-.-.-.n.a.s  a.-.p.-.-.-.n.a.s  a.-.p.-.-.-.n.n.s  a.-.p.-.-.-.n.n.s   \n",
       "\n",
       "                CL07               CL08  ...               ML10  \\\n",
       "0  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-  ...  n.-.s.-.-.-.n.g.-   \n",
       "1  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  ...  d.-.-.-.-.-.-.-.-   \n",
       "2  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  ...  r.-.-.-.-.-.-.-.-   \n",
       "3  a.-.p.-.-.-.n.n.-  a.-.p.-.-.-.n.n.-  ...  a.-.s.-.-.-.n.g.p   \n",
       "4  a.-.p.-.-.-.n.n.s  a.-.p.-.-.-.n.n.s  ...  a.-.p.-.-.-.n.a.s   \n",
       "\n",
       "               Max02              Max03              Max04              Max05  \\\n",
       "0  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-   \n",
       "1  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "2  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "3  a.-.p.-.-.-.m.b.-  a.-.s.-.-.-.f.b.p  a.-.s.-.-.-.f.b.p  a.-.s.-.-.-.f.b.p   \n",
       "4  a.-.p.-.-.-.n.a.s  a.-.s.-.-.-.f.n.s  a.-.s.-.-.-.f.n.s  a.-.s.-.-.-.f.n.s   \n",
       "\n",
       "               Max06              Max07              Max08              Max09  \\\n",
       "0  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-  n.-.s.-.-.-.n.g.-   \n",
       "1  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "2  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "3  a.-.s.-.-.-.f.b.p  a.-.s.-.-.-.f.b.p  a.-.s.-.-.-.f.b.p  a.-.s.-.-.-.f.b.p   \n",
       "4  a.-.s.-.-.-.f.n.s  a.-.s.-.-.-.f.n.s  a.-.s.-.-.-.f.n.s  a.-.s.-.-.-.f.n.s   \n",
       "\n",
       "               Max10  \n",
       "0  n.-.s.-.-.-.n.g.-  \n",
       "1  d.-.-.-.-.-.-.-.-  \n",
       "2  r.-.-.-.-.-.-.-.-  \n",
       "3  a.-.s.-.-.-.f.b.p  \n",
       "4  a.-.s.-.-.-.f.n.s  \n",
       "\n",
       "[5 rows x 39 columns]"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Ceyx_prediction_names = [\"Ceyx-predict-CL02.txt\", \"Ceyx-predict-CL03.txt\", \"Ceyx-predict-CL04.txt\", \"Ceyx-predict-CL05.txt\", \"Ceyx-predict-CL06.txt\", \"Ceyx-predict-CL07.txt\", \"Ceyx-predict-CL08.txt\", \"Ceyx-predict-CL09.txt\", \"Ceyx-predict-CL10.txt\", \"Ceyx-predict-LL02.txt\", \"Ceyx-predict-LL03.txt\", \"Ceyx-predict-LL04.txt\", \"Ceyx-predict-LL05.txt\", \"Ceyx-predict-LL06.txt\", \"Ceyx-predict-LL07.txt\", \"Ceyx-predict-LL08.txt\", \"Ceyx-predict-LL09.txt\", \"Ceyx-predict-LL10.txt\", \"Ceyx-predict-ML02.txt\", \"Ceyx-predict-ML03.txt\", \"Ceyx-predict-ML04.txt\", \"Ceyx-predict-ML05.txt\", \"Ceyx-predict-ML06.txt\", \"Ceyx-predict-ML07.txt\", \"Ceyx-predict-ML08.txt\", \"Ceyx-predict-ML09.txt\", \"Ceyx-predict-ML10.txt\", \"Ceyx-predict-Max02.txt\", \"Ceyx-predict-Max03.txt\", \"Ceyx-predict-Max04.txt\", \"Ceyx-predict-Max05.txt\", \"Ceyx-predict-Max06.txt\", \"Ceyx-predict-Max07.txt\", \"Ceyx-predict-Max08.txt\", \"Ceyx-predict-Max09.txt\", \"Ceyx-predict-Max10.txt\"]\n",
    "\n",
    "column_names = [\"Token\", \"GT Tag\", \"Lemma\", \"CL02\", \"CL03\", \"CL04\", \"CL05\", \"CL06\", \"CL07\", \"CL08\", \"CL09\", \"CL10\", \"LL02\", \"LL03\", \"LL04\", \"LL05\", \"LL06\", \"LL07\", \"LL08\", \"LL09\", \"LL10\", \"ML02\", \"ML03\", \"ML04\", \"ML05\", \"ML06\", \"ML07\", \"ML08\", \"ML09\", \"ML10\", \"Max02\", \"Max03\", \"Max04\", \"Max05\", \"Max06\", \"Max07\", \"Max08\", \"Max09\", \"Max10\"]\n",
    "df3 = pd.read_csv(\"Ceyx-gt.txt\", sep = \"\\t\", header = None, index_col = False, na_filter = False)\n",
    "\n",
    "for i in Ceyx_prediction_names:\n",
    "    df_i = pd.read_csv(f\"{i}\", sep = \"\\t\", header = None, index_col = False, na_filter = False)\n",
    "    df_i = df_i[1]\n",
    "    df3 = pd.concat([df3, df_i], axis = 1)\n",
    "df3.columns = column_names\n",
    "df3.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CL02</th>\n",
       "      <th>CL03</th>\n",
       "      <th>CL04</th>\n",
       "      <th>CL05</th>\n",
       "      <th>CL06</th>\n",
       "      <th>CL07</th>\n",
       "      <th>CL08</th>\n",
       "      <th>CL09</th>\n",
       "      <th>CL10</th>\n",
       "      <th>LL02</th>\n",
       "      <th>...</th>\n",
       "      <th>ML10</th>\n",
       "      <th>Max02</th>\n",
       "      <th>Max03</th>\n",
       "      <th>Max04</th>\n",
       "      <th>Max05</th>\n",
       "      <th>Max06</th>\n",
       "      <th>Max07</th>\n",
       "      <th>Max08</th>\n",
       "      <th>Max09</th>\n",
       "      <th>Max10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 36 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    CL02   CL03   CL04   CL05   CL06   CL07   CL08   CL09   CL10   LL02  ...  \\\n",
       "0   True   True   True   True   True   True   True   True   True   True  ...   \n",
       "1   True   True   True   True   True   True   True   True   True   True  ...   \n",
       "2   True  False  False   True   True   True   True   True   True  False  ...   \n",
       "3  False  False  False  False  False  False  False  False  False  False  ...   \n",
       "4  False  False  False  False  False  False  False  False  False  False  ...   \n",
       "\n",
       "    ML10  Max02  Max03  Max04  Max05  Max06  Max07  Max08  Max09  Max10  \n",
       "0   True   True   True   True   True   True   True   True   True   True  \n",
       "1   True   True   True   True   True   True   True   True   True   True  \n",
       "2  False  False  False  False  False  False  False  False  False  False  \n",
       "3   True  False  False  False  False  False  False  False  False  False  \n",
       "4  False  False  False  False  False  False  False  False  False  False  \n",
       "\n",
       "[5 rows x 36 columns]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df4 = pd.DataFrame()\n",
    "for column in df3.columns[3:]:\n",
    "    df4[column] = np.where(df3[column]== df3[\"GT Tag\"], True, False)\n",
    "df4.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CL02     74.7\n",
      "CL03     75.5\n",
      "CL04     75.7\n",
      "CL05     75.0\n",
      "CL06     75.2\n",
      "CL07     75.1\n",
      "CL08     75.0\n",
      "CL09     75.7\n",
      "CL10     75.4\n",
      "LL02     65.5\n",
      "LL03     65.8\n",
      "LL04     64.0\n",
      "LL05     64.3\n",
      "LL06     63.6\n",
      "LL07     63.8\n",
      "LL08     64.0\n",
      "LL09     63.6\n",
      "LL10     63.6\n",
      "ML02     70.9\n",
      "ML03     71.6\n",
      "ML04     71.3\n",
      "ML05     72.2\n",
      "ML06     71.8\n",
      "ML07     72.2\n",
      "ML08     72.0\n",
      "ML09     72.0\n",
      "ML10     72.2\n",
      "Max02    75.0\n",
      "Max03    76.1\n",
      "Max04    76.2\n",
      "Max05    75.9\n",
      "Max06    76.4\n",
      "Max07    76.2\n",
      "Max08    75.7\n",
      "Max09    76.1\n",
      "Max10    76.4\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "Ceyx_acc = df4.sum(axis = 0)/df4.shape[0]\n",
    "Ceyx_acc = Ceyx_acc.multiply(100).round(decimals = 1)\n",
    "print(Ceyx_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Token</th>\n",
       "      <th>GT Tag</th>\n",
       "      <th>Lemma</th>\n",
       "      <th>CL02</th>\n",
       "      <th>CL03</th>\n",
       "      <th>CL04</th>\n",
       "      <th>CL05</th>\n",
       "      <th>CL06</th>\n",
       "      <th>CL07</th>\n",
       "      <th>CL08</th>\n",
       "      <th>...</th>\n",
       "      <th>ML10</th>\n",
       "      <th>Max02</th>\n",
       "      <th>Max03</th>\n",
       "      <th>Max04</th>\n",
       "      <th>Max05</th>\n",
       "      <th>Max06</th>\n",
       "      <th>Max07</th>\n",
       "      <th>Max08</th>\n",
       "      <th>Max09</th>\n",
       "      <th>Max10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>non</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>non</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>...</td>\n",
       "      <td>e.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>autem</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>autem</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>d.-.-.-.-.-.-.-.-</td>\n",
       "      <td>...</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "      <td>c.-.-.-.-.-.-.-.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>a</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>ab</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>...</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "      <td>r.-.-.-.-.-.-.-.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>seipso</td>\n",
       "      <td>n.-.s.-.-.-.m.b.-</td>\n",
       "      <td>seipse</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>n.-.s.-.-.-.n.b.-</td>\n",
       "      <td>a.-.s.-.-.-.n.b.p</td>\n",
       "      <td>a.-.s.-.-.-.n.b.p</td>\n",
       "      <td>a.-.s.-.-.-.n.b.p</td>\n",
       "      <td>a.-.s.-.-.-.n.b.p</td>\n",
       "      <td>...</td>\n",
       "      <td>p.-.s.-.-.-.m.b.-</td>\n",
       "      <td>p.-.s.-.-.-.m.b.-</td>\n",
       "      <td>p.-.s.-.-.-.n.b.-</td>\n",
       "      <td>p.-.s.-.-.-.n.b.-</td>\n",
       "      <td>p.-.s.-.-.-.n.b.-</td>\n",
       "      <td>p.-.s.-.-.-.n.b.-</td>\n",
       "      <td>p.-.s.-.-.-.n.b.-</td>\n",
       "      <td>p.-.s.-.-.-.n.b.-</td>\n",
       "      <td>p.-.s.-.-.-.n.b.-</td>\n",
       "      <td>p.-.s.-.-.-.n.b.-</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>:</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>:</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>...</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "      <td>u.-.-.-.-.-.-.-.-</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 39 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    Token             GT Tag   Lemma               CL02               CL03  \\\n",
       "0     non  d.-.-.-.-.-.-.-.-     non  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "1   autem  c.-.-.-.-.-.-.-.-   autem  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "2       a  r.-.-.-.-.-.-.-.-      ab  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "3  seipso  n.-.s.-.-.-.m.b.-  seipse  n.-.s.-.-.-.n.b.-  n.-.s.-.-.-.n.b.-   \n",
       "4       :  u.-.-.-.-.-.-.-.-       :  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-   \n",
       "\n",
       "                CL04               CL05               CL06               CL07  \\\n",
       "0  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "1  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "2  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "3  n.-.s.-.-.-.n.b.-  a.-.s.-.-.-.n.b.p  a.-.s.-.-.-.n.b.p  a.-.s.-.-.-.n.b.p   \n",
       "4  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-   \n",
       "\n",
       "                CL08  ...               ML10              Max02  \\\n",
       "0  d.-.-.-.-.-.-.-.-  ...  e.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "1  d.-.-.-.-.-.-.-.-  ...  c.-.-.-.-.-.-.-.-  c.-.-.-.-.-.-.-.-   \n",
       "2  r.-.-.-.-.-.-.-.-  ...  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "3  a.-.s.-.-.-.n.b.p  ...  p.-.s.-.-.-.m.b.-  p.-.s.-.-.-.m.b.-   \n",
       "4  u.-.-.-.-.-.-.-.-  ...  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-   \n",
       "\n",
       "               Max03              Max04              Max05              Max06  \\\n",
       "0  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-   \n",
       "1  c.-.-.-.-.-.-.-.-  c.-.-.-.-.-.-.-.-  c.-.-.-.-.-.-.-.-  c.-.-.-.-.-.-.-.-   \n",
       "2  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-   \n",
       "3  p.-.s.-.-.-.n.b.-  p.-.s.-.-.-.n.b.-  p.-.s.-.-.-.n.b.-  p.-.s.-.-.-.n.b.-   \n",
       "4  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-   \n",
       "\n",
       "               Max07              Max08              Max09              Max10  \n",
       "0  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  d.-.-.-.-.-.-.-.-  \n",
       "1  c.-.-.-.-.-.-.-.-  c.-.-.-.-.-.-.-.-  c.-.-.-.-.-.-.-.-  c.-.-.-.-.-.-.-.-  \n",
       "2  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  r.-.-.-.-.-.-.-.-  \n",
       "3  p.-.s.-.-.-.n.b.-  p.-.s.-.-.-.n.b.-  p.-.s.-.-.-.n.b.-  p.-.s.-.-.-.n.b.-  \n",
       "4  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-  u.-.-.-.-.-.-.-.-  \n",
       "\n",
       "[5 rows x 39 columns]"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "IT_prediction_names = [\"IT-predict-CL02.txt\", \"IT-predict-CL03.txt\", \"IT-predict-CL04.txt\", \"IT-predict-CL05.txt\", \"IT-predict-CL06.txt\", \"IT-predict-CL07.txt\", \"IT-predict-CL08.txt\", \"IT-predict-CL09.txt\", \"IT-predict-CL10.txt\", \"IT-predict-LL02.txt\", \"IT-predict-LL03.txt\", \"IT-predict-LL04.txt\", \"IT-predict-LL05.txt\", \"IT-predict-LL06.txt\", \"IT-predict-LL07.txt\", \"IT-predict-LL08.txt\", \"IT-predict-LL09.txt\", \"IT-predict-LL10.txt\", \"IT-predict-ML02.txt\", \"IT-predict-ML03.txt\", \"IT-predict-ML04.txt\", \"IT-predict-ML05.txt\", \"IT-predict-ML06.txt\", \"IT-predict-ML07.txt\", \"IT-predict-ML08.txt\", \"IT-predict-ML09.txt\", \"IT-predict-ML10.txt\", \"IT-predict-Max02.txt\", \"IT-predict-Max03.txt\", \"IT-predict-Max04.txt\", \"IT-predict-Max05.txt\", \"IT-predict-Max06.txt\", \"IT-predict-Max07.txt\", \"IT-predict-Max08.txt\", \"IT-predict-Max09.txt\", \"IT-predict-Max10.txt\"]\n",
    "column_names = [\"Token\", \"GT Tag\", \"Lemma\", \"CL02\", \"CL03\", \"CL04\", \"CL05\", \"CL06\", \"CL07\", \"CL08\", \"CL09\", \"CL10\", \"LL02\", \"LL03\", \"LL04\", \"LL05\", \"LL06\", \"LL07\", \"LL08\", \"LL09\", \"LL10\", \"ML02\", \"ML03\", \"ML04\", \"ML05\", \"ML06\", \"ML07\", \"ML08\", \"ML09\", \"ML10\", \"Max02\", \"Max03\", \"Max04\", \"Max05\", \"Max06\", \"Max07\", \"Max08\", \"Max09\", \"Max10\"]\n",
    "df5 = pd.read_csv(\"IT-test-gt.txt\", sep = \"\\t\", header = None, index_col = False, na_filter = False)\n",
    "\n",
    "for i in IT_prediction_names:\n",
    "    df_i = pd.read_csv(f\"{i}\", sep = \"\\t\", header = None, index_col = False, na_filter = False)\n",
    "    df_i = df_i[1]\n",
    "    df5 = pd.concat([df5, df_i], axis = 1)\n",
    "df5.columns = column_names\n",
    "df5.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>CL02</th>\n",
       "      <th>CL03</th>\n",
       "      <th>CL04</th>\n",
       "      <th>CL05</th>\n",
       "      <th>CL06</th>\n",
       "      <th>CL07</th>\n",
       "      <th>CL08</th>\n",
       "      <th>CL09</th>\n",
       "      <th>CL10</th>\n",
       "      <th>LL02</th>\n",
       "      <th>...</th>\n",
       "      <th>ML10</th>\n",
       "      <th>Max02</th>\n",
       "      <th>Max03</th>\n",
       "      <th>Max04</th>\n",
       "      <th>Max05</th>\n",
       "      <th>Max06</th>\n",
       "      <th>Max07</th>\n",
       "      <th>Max08</th>\n",
       "      <th>Max09</th>\n",
       "      <th>Max10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 36 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    CL02   CL03   CL04   CL05   CL06   CL07   CL08   CL09   CL10   LL02  ...  \\\n",
       "0   True   True   True   True   True   True   True   True   True   True  ...   \n",
       "1  False  False  False  False  False  False  False  False  False  False  ...   \n",
       "2   True   True   True   True   True   True   True   True   True   True  ...   \n",
       "3  False  False  False  False  False  False  False  False  False  False  ...   \n",
       "4   True   True   True   True   True   True   True   True   True   True  ...   \n",
       "\n",
       "    ML10  Max02  Max03  Max04  Max05  Max06  Max07  Max08  Max09  Max10  \n",
       "0  False   True   True   True   True   True   True   True   True   True  \n",
       "1   True   True   True   True   True   True   True   True   True   True  \n",
       "2   True   True   True   True   True   True   True   True   True   True  \n",
       "3  False  False  False  False  False  False  False  False  False  False  \n",
       "4   True   True   True   True   True   True   True   True   True   True  \n",
       "\n",
       "[5 rows x 36 columns]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df6 = pd.DataFrame()\n",
    "for column in df5.columns[3:]:\n",
    "    df6[column] = np.where(df5[column]== df5[\"GT Tag\"], True, False)\n",
    "df6.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CL02     70.1\n",
      "CL03     70.3\n",
      "CL04     70.3\n",
      "CL05     70.3\n",
      "CL06     70.2\n",
      "CL07     70.3\n",
      "CL08     70.3\n",
      "CL09     70.2\n",
      "CL10     70.2\n",
      "LL02     66.6\n",
      "LL03     66.5\n",
      "LL04     66.3\n",
      "LL05     66.2\n",
      "LL06     66.3\n",
      "LL07     66.3\n",
      "LL08     66.3\n",
      "LL09     66.2\n",
      "LL10     66.2\n",
      "ML02     89.0\n",
      "ML03     89.0\n",
      "ML04     89.0\n",
      "ML05     89.0\n",
      "ML06     89.0\n",
      "ML07     89.1\n",
      "ML08     89.0\n",
      "ML09     89.0\n",
      "ML10     89.0\n",
      "Max02    87.2\n",
      "Max03    87.4\n",
      "Max04    87.4\n",
      "Max05    87.4\n",
      "Max06    87.4\n",
      "Max07    87.3\n",
      "Max08    87.3\n",
      "Max09    87.2\n",
      "Max10    87.2\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "IT_acc = df6.sum(axis = 0)/df6.shape[0]\n",
    "IT_acc = IT_acc.multiply(100).round(decimals = 1)\n",
    "\n",
    "print(IT_acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Figure 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.style.use('bmh')\n",
    "plt.rcParams[\"font.family\"] = \"Arial\"\n",
    "\n",
    "fig, (ax1, ax2, ax3) = plt.subplots(3, figsize=(10, 10), sharex=True)\n",
    "\n",
    "y_label = [2,3,4,5,6,7,8,9,10]\n",
    "\n",
    "ax1.plot(y_label, Boudica_acc[0:9], \"black\")\n",
    "ax1.plot(y_label, Boudica_acc[9:18], \"darkred\") \n",
    "ax1.plot(y_label, Boudica_acc[18:27], \"blue\")\n",
    "ax1.plot(y_label, Boudica_acc[27:], \"green\")\n",
    "ax1.set_ylabel(\"Accuracy (%)\")\n",
    "ax1.legend([\"CL\", \"LL\", \"ML\", \"Max\"],bbox_to_anchor=(1.05, 1), loc='upper left', borderaxespad=0.)\n",
    "ax1.set_title(\"Boudica Test Set\")\n",
    "ax1.set_xlabel(\"n-gram Context Size\")\n",
    "\n",
    "ax2.plot(y_label, Ceyx_acc[0:9], \"black\")\n",
    "ax2.plot(y_label, Ceyx_acc[9:18], \"darkred\")\n",
    "ax2.plot(y_label, Ceyx_acc[18:27], \"blue\")\n",
    "ax2.plot(y_label, Ceyx_acc[27:], \"green\")\n",
    "ax2.set_ylabel(\"Accuracy (%)\")\n",
    "ax2.set_title(\"Ceyx Test Set\")\n",
    "ax2.set_xlabel(\"n-gram Context Size\")\n",
    "\n",
    "ax3.plot(y_label, IT_acc[0:9], \"black\")\n",
    "ax3.plot(y_label, IT_acc[9:18], \"darkred\")\n",
    "ax3.plot(y_label, IT_acc[18:27], \"blue\")\n",
    "ax3.plot(y_label, IT_acc[27:], \"green\")\n",
    "ax3.set_ylabel(\"Accuracy (%)\")\n",
    "ax3.set_title(\"Index Thomisticus Test Set\")\n",
    "ax3.set_xlabel(\"n-gram Context Size\")\n",
    "\n",
    "fig.tight_layout()\n",
    "\n",
    "plt.savefig(\"DHTP01.pdf\")\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}