plot-hop1-hop2-last-hop-contribution.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook plots the private network latency contribution as a ratio of hop1 to hop2 latencies.\n",
    "- - - "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: pylab import has clobbered these variables: ['yticks', 'f']\n",
      "`%matplotlib` prevents importing * from pylab and numpy\n"
     ]
    }
   ],
   "source": [
    "%pylab inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>body {\n",
       "    margin: 0;\n",
       "    font-family: Helvetica;\n",
       "}\n",
       "table.dataframe {\n",
       "    border-collapse: collapse;\n",
       "    border: none;\n",
       "}\n",
       "table.dataframe tr {\n",
       "    border: none;\n",
       "}\n",
       "table.dataframe td, table.dataframe th {\n",
       "    margin: 0;\n",
       "    border: 1px solid white;\n",
       "    padding-left: 0.25em;\n",
       "    padding-right: 0.25em;\n",
       "}\n",
       "table.dataframe th:not(:empty) {\n",
       "    background-color: #fec;\n",
       "    text-align: left;\n",
       "    font-weight: normal;\n",
       "}\n",
       "table.dataframe tr:nth-child(2) th:empty {\n",
       "    border-left: none;\n",
       "    border-right: 1px dashed #888;\n",
       "}\n",
       "table.dataframe td {\n",
       "    border: 2px solid #ccf;\n",
       "    background-color: #f4f4ff;\n",
       "}\n",
       "h3 {\n",
       "    color: white;\n",
       "    background-color: black;\n",
       "    padding: 0.5em;\n",
       "}\n",
       "\n",
       ".CodeMirror {\n",
       "    font-family: 'Monaco for Powerline';\n",
       "}\n",
       "\n",
       "div.cell.code_cell {\n",
       "    font-family: 'Monaco for Powerline';\n",
       "}\n",
       "\n",
       "div.output_area pre {\n",
       "    font-family: 'Monaco for Powerline';\n",
       "}\n",
       "</style>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.core.display import HTML\n",
    "css = open('style-table.css').read() + open('style-notebook.css').read()\n",
    "HTML('<style>{}</style>'.format(css))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import sqlite3\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "DB_LOCATION = 'lastmile.db'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "RA_PROBE = 'ra_probe'\n",
    "RA_ONE_OFF_EXTENSIONS = 'ra_one_off_extensions'\n",
    "RA_LATENCIES_AGG = 'ra_latencies_agg'\n",
    "RA_LATENCIES_LAST_HOP_AGG = 'ra_latencies_last_hop_agg'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "SK_ONE_OFF_EXTENSIONS = 'sk_one_off_extensions'\n",
    "SK_LATENCIES_AGG = 'sk_latencies_agg'\n",
    "SK_LATENCIES_LAST_HOP_AGG = 'sk_latencies_last_hop_agg'\n",
    "SK_ASN = 'sk_asn'\n",
    "METADATA_SNAPSHOT = 'metadata_snapshot'\n",
    "METADATA_SNAPSHOT_EXTENSIONS = 'metadata_snapshot_extensions'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "con = sqlite3.connect(DB_LOCATION)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "query_ra = '''SELECT      agg.probeid\n",
    "                        , agg.timestamp\n",
    "                        , agg.h1_median || ';' || agg.h2_median || ';' || lh.last_hop_median as h1h2lh\n",
    "\n",
    "               FROM   %s                             as  agg\n",
    "                    , %s                             as  hardware\n",
    "                    , %s                             as  ext\n",
    "                    , %s                             as  lh\n",
    "\n",
    "               WHERE     agg.probeid                  =  hardware.probeid\n",
    "               AND       agg.probeid                  =  ext.probeid\n",
    "               AND       agg.probeid                  =  lh.probeid\n",
    "               AND       agg.timestamp                =  lh.timestamp               \n",
    "\n",
    "               AND       hardware.hardware_version    =  'v3'\n",
    "               AND       ext.if_residential           =  1\n",
    "               \n",
    "            '''%(  RA_LATENCIES_AGG\n",
    "                 , RA_PROBE\n",
    "                 , RA_ONE_OFF_EXTENSIONS\n",
    "                 , RA_LATENCIES_LAST_HOP_AGG\n",
    "                )\n",
    "\n",
    "df_ra = pd.read_sql(query_ra, con)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "probeid      133533\n",
       "timestamp    133533\n",
       "h1h2lh       129045\n",
       "dtype: int64"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_ra.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>probeid</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>h1h2lh</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10006</td>\n",
       "      <td>1406548034</td>\n",
       "      <td>0.429;20.49;64.929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10006</td>\n",
       "      <td>1406562432</td>\n",
       "      <td>0.431;17.7;64.194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10006</td>\n",
       "      <td>1406576835</td>\n",
       "      <td>0.438;17.776;64.473</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10006</td>\n",
       "      <td>1406591230</td>\n",
       "      <td>0.478;17.577;64.46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10006</td>\n",
       "      <td>1406605631</td>\n",
       "      <td>0.422;17.526;64.059</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   probeid   timestamp               h1h2lh\n",
       "0    10006  1406548034   0.429;20.49;64.929\n",
       "1    10006  1406562432    0.431;17.7;64.194\n",
       "2    10006  1406576835  0.438;17.776;64.473\n",
       "3    10006  1406591230   0.478;17.577;64.46\n",
       "4    10006  1406605631  0.422;17.526;64.059"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_ra.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "query_sk = ''' SELECT     agg.probeid\n",
    "                        , agg.timestamp\n",
    "                        , s.asn       \n",
    "                        , s.asn_holder_name\n",
    "                        , meta.product                        \n",
    "                        , agg.h1_avg || ';' || agg.h2_avg || ';' || lh.last_hop_avg as h1h2lh\n",
    "\n",
    "                                                                                     \n",
    "               FROM       %s                            as   agg\n",
    "                        , %s                            as   api               \n",
    "                        , %s                            as   ext\n",
    "                        , %s                            as   s\n",
    "                        , %s                            as   meta\n",
    "                        , %s                            as   lh\n",
    "                        \n",
    "                        \n",
    "\n",
    "               WHERE      agg.probeid                    =    ext.probeid\n",
    "               AND        agg.probeid                    =    api.probeid\n",
    "               AND        agg.probeid                    =    meta.id               \n",
    "               AND        agg.probeid                    =    lh.probeid\n",
    "               AND        agg.timestamp                  =    lh.timestamp\n",
    "               AND        api.asn                        =    s.asn               \n",
    "               \n",
    "               AND        ext.if_residential             =    1\n",
    "               AND        agg.h1_avg                     <>   0\n",
    "               AND        agg.h2_avg                     <>   0\n",
    "               AND        lh.last_hop_avg                <>   0               \n",
    "               \n",
    "           '''%(   SK_LATENCIES_AGG\n",
    "                 , METADATA_SNAPSHOT_EXTENSIONS                \n",
    "                 , SK_ONE_OFF_EXTENSIONS\n",
    "                 , SK_ASN        \n",
    "                 , METADATA_SNAPSHOT                \n",
    "                 , SK_LATENCIES_LAST_HOP_AGG\n",
    "               )\n",
    "\n",
    "df_sk = pd.read_sql(query_sk, con)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "probeid            435813\n",
       "timestamp          435813\n",
       "asn                435813\n",
       "asn_holder_name    435813\n",
       "product            435813\n",
       "h1h2lh             435813\n",
       "dtype: int64"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sk.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>probeid</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>asn</th>\n",
       "      <th>asn_holder_name</th>\n",
       "      <th>product</th>\n",
       "      <th>h1h2lh</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>14905</td>\n",
       "      <td>2014-07-28 00:36:14</td>\n",
       "      <td>2856</td>\n",
       "      <td>BT-UK-AS BT Public Internet Service,GB</td>\n",
       "      <td>40</td>\n",
       "      <td>0.794;7.423;22.676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>14905</td>\n",
       "      <td>2014-07-28 00:36:18</td>\n",
       "      <td>2856</td>\n",
       "      <td>BT-UK-AS BT Public Internet Service,GB</td>\n",
       "      <td>40</td>\n",
       "      <td>0.787;7.41;100.878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>14905</td>\n",
       "      <td>2014-07-28 12:36:52</td>\n",
       "      <td>2856</td>\n",
       "      <td>BT-UK-AS BT Public Internet Service,GB</td>\n",
       "      <td>40</td>\n",
       "      <td>0.786;7.687;24.642</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>14905</td>\n",
       "      <td>2014-07-28 12:36:56</td>\n",
       "      <td>2856</td>\n",
       "      <td>BT-UK-AS BT Public Internet Service,GB</td>\n",
       "      <td>40</td>\n",
       "      <td>0.808;7.539;98.557</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>14905</td>\n",
       "      <td>2014-07-28 16:36:10</td>\n",
       "      <td>2856</td>\n",
       "      <td>BT-UK-AS BT Public Internet Service,GB</td>\n",
       "      <td>40</td>\n",
       "      <td>0.813;7.576;19.646</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   probeid            timestamp   asn                         asn_holder_name  \\\n",
       "0    14905  2014-07-28 00:36:14  2856  BT-UK-AS BT Public Internet Service,GB   \n",
       "1    14905  2014-07-28 00:36:18  2856  BT-UK-AS BT Public Internet Service,GB   \n",
       "2    14905  2014-07-28 12:36:52  2856  BT-UK-AS BT Public Internet Service,GB   \n",
       "3    14905  2014-07-28 12:36:56  2856  BT-UK-AS BT Public Internet Service,GB   \n",
       "4    14905  2014-07-28 16:36:10  2856  BT-UK-AS BT Public Internet Service,GB   \n",
       "\n",
       "  product              h1h2lh  \n",
       "0      40  0.794;7.423;22.676  \n",
       "1      40  0.787;7.41;100.878  \n",
       "2      40  0.786;7.687;24.642  \n",
       "3      40  0.808;7.539;98.557  \n",
       "4      40  0.813;7.576;19.646  "
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sk.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_ra = df_ra.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_sk = df_sk.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "probeid            435813\n",
       "timestamp          435813\n",
       "asn                435813\n",
       "asn_holder_name    435813\n",
       "product            435813\n",
       "h1h2lh             435813\n",
       "dtype: int64"
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sk.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "probeid      129045\n",
       "timestamp    129045\n",
       "h1h2lh       129045\n",
       "dtype: int64"
      ]
     },
     "execution_count": 118,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_ra.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def calculate_private_contribution(h1h2lh):\n",
    "    try: h1,h2,lh = h1h2lh.split(';')\n",
    "    except Exception as e: print(e, type(e), h1h2); return None\n",
    "    \n",
    "    try: contribution = float(h1) / float(h2) * 100\n",
    "    except Exception as e: print(e, type(e), h1, h2); return None\n",
    "    \n",
    "    return contribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def calculate_contribution(h1h2lh):\n",
    "    try: h1,h2,lh = h1h2lh.split(';')\n",
    "    except Exception as e: print(e, type(e), h1h2); return None\n",
    "    \n",
    "    try: contribution = float(h2) - float(h1) / float(lh) * 100\n",
    "    except Exception as e: print(e, type(e), h1, h2); return None\n",
    "    \n",
    "    return contribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_ra['lm_contribution'] = df_ra['h1h2lh'].apply(calculate_contribution)\n",
    "df_sk['lm_contribution'] = df_sk['h1h2lh'].apply(calculate_contribution)\n",
    "\n",
    "df_ra['pv_contribution'] = df_ra['h1h2lh'].apply(calculate_private_contribution)\n",
    "df_sk['pv_contribution'] = df_sk['h1h2lh'].apply(calculate_private_contribution)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_ra_groupby = df_ra.groupby('probeid')\n",
    "df_ra_agg = df_ra_groupby['lm_contribution', 'pv_contribution'].agg(['median'])\n",
    "df_ra_merged = pd.merge(df_ra, df_ra_agg, left_on='probeid', right_index=True)\n",
    "\n",
    "del df_ra_merged['timestamp']\n",
    "del df_ra_merged['h1h2lh']\n",
    "del df_ra_merged['lm_contribution']\n",
    "del df_ra_merged['pv_contribution']\n",
    "\n",
    "df_ra_merged = df_ra_merged.drop_duplicates()\n",
    "df_ra_merged.rename(columns={'median': 'lm_contribution'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_sk_groupby = df_sk.groupby('probeid')\n",
    "df_sk_agg = df_sk_groupby['lm_contribution', 'pv_contribution'].agg(['median'])\n",
    "df_sk_merged = pd.merge(df_sk, df_sk_agg, left_on='probeid', right_index=True)\n",
    "\n",
    "del df_sk_merged['timestamp']\n",
    "del df_sk_merged['h1h2lh']\n",
    "del df_sk_merged['lm_contribution']\n",
    "del df_sk_merged['pv_contribution']\n",
    "\n",
    "df_sk_merged = df_sk_merged.drop_duplicates()\n",
    "df_sk_merged.rename(columns={'median': 'lm_contribution'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "probeid                      691\n",
       "(lm_contribution, median)    691\n",
       "(pv_contribution, median)    691\n",
       "dtype: int64"
      ]
     },
     "execution_count": 124,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_ra_merged.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "probeid                      1245\n",
       "asn                          1245\n",
       "asn_holder_name              1245\n",
       "product                      1245\n",
       "(lm_contribution, median)    1245\n",
       "(pv_contribution, median)    1245\n",
       "dtype: int64"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sk_merged.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_sk_merged.columns = ['probeid', 'asn', 'asn_holder_name', 'product', 'lm_contribution', 'pv_contribution']\n",
    "df_ra_merged.columns = ['probeid', 'lm_contribution', 'pv_contribution']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "probeid            3\n",
       "lm_contribution    3\n",
       "pv_contribution    3\n",
       "dtype: int64"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_ra_merged[df_ra_merged.pv_contribution >= 100].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "probeid            124\n",
       "asn                124\n",
       "asn_holder_name    124\n",
       "product            124\n",
       "lm_contribution    124\n",
       "pv_contribution    124\n",
       "dtype: int64"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sk_merged[df_sk_merged.pv_contribution >= 100].count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### remove probes where hop1 > 100% of hop2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_sk_merged = df_sk_merged[df_sk_merged.pv_contribution < 100]\n",
    "df_ra_merged = df_ra_merged[df_ra_merged.pv_contribution < 100]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import Cdf\n",
    "import Pmf\n",
    "ra_sample = df_ra_merged['lm_contribution'].values\n",
    "sk_sample = df_sk_merged['lm_contribution'].values\n",
    "ra_c = Cdf.MakeCdfFromList(ra_sample)\n",
    "sk_c = Cdf.MakeCdfFromList(sk_sample)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWwAAACiCAYAAABlCvfJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8s+b/gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7ff9bd553a90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# ---- plt.style.use('ggplot')\n",
    "fig = plt.figure(figsize=(5,1.5))\n",
    "plt.rcParams['font.size'] = 12\n",
    "yticks = np.arange(0,1.1,0.2)\n",
    "ax = plt.gca()\n",
    "\n",
    "ax.plot (   sk_c.xs\n",
    "          , sk_c.ps\n",
    "          , marker='s'\n",
    "          , linewidth=0.3\n",
    "          , markersize=5\n",
    "          , fillstyle='none'\n",
    "          , color='Blue'\n",
    "         )\n",
    "\n",
    "ax.plot (  ra_c.xs\n",
    "          , ra_c.ps\n",
    "          , marker='x'\n",
    "          , linewidth=0.3\n",
    "          , markersize=5\n",
    "          , color='Violet'\n",
    "         )\n",
    "\n",
    "ax.spines['right'].set_color('none')\n",
    "ax.spines['top'].set_color('none')\n",
    "ax.yaxis.set_ticks_position('left')\n",
    "ax.xaxis.set_ticks_position('bottom')\n",
    "ax.spines['bottom'].set_position(('axes', -0.05))\n",
    "ax.spines['left'].set_position(('axes', -0.05))\n",
    "\n",
    "ax.yaxis.set_label_coords(-0.21, 0.45)\n",
    "ax = fig.axes[0]\n",
    "ax.grid(False)\n",
    "ax.set_yticks(yticks)\n",
    "ax.set_ylim([0,1.05])\n",
    "ax.set_xlim(xmin=0.1)\n",
    "ax.set_xscale('log')\n",
    "ax.set_xlabel(' (hop2 - hop1) / last-hop (%)', fontsize=12)\n",
    "ax.set_ylabel('CDF', fontsize=12)\n",
    "ax.legend(['SamKnows   (1.1K)', 'RIPE Atlas (0.6K)'], fontsize=12, bbox_to_anchor=(0.9, -0.55))\n",
    "ax.legend(['SamKnows   (1.1K)', 'RIPE Atlas (0.6K)'], fontsize=12, loc='best')\n",
    "plt.title('Residential Probes', fontsize=12, y=1.05)\n",
    "plt.savefig('ripeatlas-samknows-lm-contribution.pdf', format='pdf', bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "with open('ra-lm-contribution.csv', 'w') as f: \n",
    "    f.write('values; percentile\\n')\n",
    "    for x, y in zip(ra_c.xs, ra_c.ps): f.write('%.2f; %.2f\\n'%(x,y))\n",
    "        \n",
    "with open('sk-lm-contribution.csv', 'w') as f: \n",
    "    f.write('values; percentile\\n')\n",
    "    for x, y in zip(sk_c.xs, sk_c.ps): f.write('%.2f; %.2f\\n'%(x,y))        "
   ]
  }
 ],
 "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": "3.4.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}