Dataharvest-PythonIV-2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This tutorial will show you how to load and manipulate files in Python. The key take-away here is the power of scripting. By scripting the file processing, you can batch import many files at once, you can clean-up and add values to your data, and you can repeat the task later when you get new files."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We are going to process a large file from the U.S. Bureau of Labor Statistics, the 2016 annual report on employment for the entire U.S., and all U.S. states and counties, broken down by industry and ownership sector (public employers and private employers).\n",
    "\n",
    "Our goal is to loop through the rows in this large file (500 MB unzipped) and filter only the rows with data for the U.S., States and counties -- we are going to skip data for metropolitan areas and other niche geographies. \n",
    "\n",
    "We are also going to filter the detail level - the source data breaks industries down six ways, depending on how detailed we want to get -- for our purposes, the grand total and the mid-level detail (three-digit industry codes) is fine.\n",
    "\n",
    "Because the industry and geographic area are provided to us as codes, we are going to create a lookup function that will insert columns contained the name of the industry and the geography. \n",
    "\n",
    "And we are also going to create a function that will attach an 'inflator' in case we want to adjust the dollar amounts for inflation -- in this case, going from 2016 to 2017 isn't such a big deal, remember, this script could be used on other years, and the more you go back in the past, the more inflation is a factor."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To start with, let's create the look-up functions that we will later call when we're doing the actuall file processing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "create a lookup for the industry title\n",
    " download this file \n",
    "https://raw.githubusercontent.com/gebelo/training2018/master/industry_titles.csv \n",
    "and save it locally... \n",
    "in theory, we could have done this step also in Python but doing it this way \n",
    "allows you to peruse the file before we work with it.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<_csv.reader at 0x10e39c6e0>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import csv\n",
    "# we're going to open the file and turn it into a Python object \n",
    "infile=open(\"industry_titles.csv\") #CHANGE THIS TO THE PATH TO YOUR FILE\n",
    "# we're going to then convert the object into a Python CSV object, which will give us access to the CSV library of commands\n",
    "reader=csv.reader(infile)\n",
    "reader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['industry_code', 'industry_title']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# we're going to start at the top of the file and declare the first row to be the headers\n",
    "headers=reader.next()\n",
    "headers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function close>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# we're going to create a Python dictionary -- a key, and a value -- that will be our lookup.\n",
    "lk_naics={}\n",
    "#we're going to loop through the remaining rows in the file and populate our dictionary\n",
    "#as we go through the row, the CSV reader is parsing each into two parts, row[0] and row[1]\n",
    "for row in reader:\n",
    "    #what we're doing here is populating lk_naics and setting the first value, row[0] to equal the second value row[1] - but we're getting rid\n",
    "    #of the string 'NAICS' and the code itself, which would be redundant. And finally, we're getting rid of any\n",
    "    #extra spaces\n",
    "    lk_naics[row[0]]=row[1].replace('NAICS','').replace(row[0],'').strip()\n",
    "#when we're done, we close the file to cleanup memory    \n",
    "infile.close"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Other waste collection\n",
      "Soybean farming\n"
     ]
    }
   ],
   "source": [
    "#these are the kind of lookups we can use when we do the main data import...\n",
    "print lk_naics[\"562119\"]\n",
    "print lk_naics[\"111110\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function close>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#the entire function\n",
    "# we're going to use the Python csv library\n",
    "import csv\n",
    "# we're going to open the file and turn it into a Python object \n",
    "infile=open(\"industry_titles.csv\")\n",
    "# we're going to then convert the object into a Python CSV object, which will give us access to the CSV library of commands\n",
    "reader=csv.reader(infile)\n",
    "# we're going to start at the top of the file and declare the first row to be the headers\n",
    "headers=reader.next()\n",
    "# we're going to create a Python dictionary -- a key, and a value -- that will be our lookup.\n",
    "lk_naics={}\n",
    "#we're going to loop through the remaining rows in the file and populate our dictionary\n",
    "#as we go through the row, the CSV reader is parsing each into two parts, row[0] and row[1]\n",
    "for row in reader:\n",
    "    #what we're doing here is populating lk_naics and setting the first value, row[0] to equal the second value row[1] - but we're getting rid\n",
    "    #of the string 'NAICS' and the code itself, which would be redundant. And finally, we're getting rid of any\n",
    "    #extra spaces\n",
    "    lk_naics[row[0]]=row[1].replace('NAICS','').replace(row[0],'').strip()\n",
    "#when we're done, we close the file to cleanup memory    \n",
    "infile.close"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bergen County, New Jersey\n"
     ]
    }
   ],
   "source": [
    "#the area titles function is more straightforward\n",
    "#download the file from here https://raw.githubusercontent.com/gebelo/training2018/master/area_titles.csv and\n",
    "#save it locally\n",
    "\n",
    "infile=open(\"area_titles.csv\") #CHANGE TO YOUR PATH\n",
    "reader=csv.reader(infile)\n",
    "headers=reader.next()\n",
    "lk_area={}\n",
    "for row in reader:\n",
    "    lk_area[row[0]]=row[1]\n",
    "    \n",
    "infile.close\n",
    "\n",
    "print lk_area[\"34003\"]\n",
    "\n",
    "#let's discuss briefly why this worked...."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.0213011\n"
     ]
    }
   ],
   "source": [
    "#and lastly, our inflation table\n",
    "#grab it from here https://raw.githubusercontent.com/gebelo/training2018/master/inflation.csv\n",
    "#note that there are different inflation indexes in here, we're using the one called \"inflator\"\n",
    "#which is column index number 2 -- the third column when we start counting with zero\n",
    "\n",
    "infile=open(\"inflation.csv\")\n",
    "reader=csv.reader(infile)\n",
    "headers=reader.next()\n",
    "lk_inflation={}\n",
    "for row in reader:\n",
    "    lk_inflation[row[0]]=row[2]\n",
    "\n",
    "\n",
    "infile.close\n",
    "print lk_inflation[\"2016\"]\n",
    "#any questions on why this worked?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " and now for the exciting conclusion of this exercise... let's grab the giant data file \n",
    "from the bls web site  -- https://data.bls.gov/cew/data/files/2016/csv/2016_annual_singlefile.zip\n",
    "unzip the file locally on your computer.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "let's look at some file layouts and code sheet to see what we're doing\n",
    "\n",
    "the file layout is here:\n",
    "https://data.bls.gov/cew/doc/layouts/csv_annual_layout.htm\n",
    "\n",
    "it looks like the key variables are columns 1 through 16 -- this means we're only going to import\n",
    "python index columns 0 through 15, and then attach our lookup values.\n",
    "\n",
    "to filter for just the total and three-digit industry codes, we refer to the table here:\n",
    "https://data.bls.gov/cew/doc/titles/agglevel/agglevel_titles.htm\n",
    "\n",
    "for the us, we want levels 10 and 15; for states, 50 and 55; for counties, we want 70 and 75\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/Users/robertgebeloff/Downloads/2016.annual.singlefile.csv'"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# create a variable with the path to the file\n",
    "f='/Users/robertgebeloff/Downloads/2016.annual.singlefile.csv' #CHANGE TO YOUR PATH\n",
    "f"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#create new file with the prefix 'new_' attached to the current file's name\n",
    "file_out = open(\"/Users/robertgebeloff/Downloads/new_2016.annual.singlefile.csv\", 'w')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#open original file    \n",
    "o = open(f)\n",
    "#create a Python csv object called 'file_in'\n",
    "file_in = csv.reader((o), delimiter=\",\")\n",
    "#create a Python object that will write to 'file-out'\n",
    "csv_writer = csv.writer(file_out)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#insert headers from original file into the new file, making room for 3 new columns based on our lookups\n",
    "csv_writer.writerow(next(file_in)+[\"area_label\",\"industry_label\",\"inflator\"])\n",
    "#loop through the original file, lookup labels and append them to row    \n",
    "for row in file_in:\n",
    "        #run our lookups and store reponse in variables\n",
    "        \n",
    "        the_area=lk_area[row[0]] # this runs 'area_fips' through our lk_area dictionary\n",
    "        the_industry=lk_naics[row[2]] # this runs 'industry_code' through our lk_naics dictionary\n",
    "        the_inflator=lk_inflation[row[5]] # this runs the year for our data through our inflation dictionary\n",
    "        row=row[0:15]\n",
    "        row.append(the_area)\n",
    "        row.append(the_industry)\n",
    "        row.append(the_inflator)\n",
    "        \n",
    "#write the row with the appendages to the new file only if it's the grand total or a three-digit industry\n",
    "        if row[3] in (\"10\",\"15\",\"50\",\"55\",\"70\",\"75\"):\n",
    "                csv_writer.writerow(row)\n",
    "\n",
    "file_out.close()\n",
    "o.close()\n",
    "\n",
    "#we've gone from 500 mbs to 40 in a few seconds...."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's do one more example -- this time we're going to do some processing on a tab-delimited file. But in addition, we're going to use some string functions to further split apart one of the fields into components.\n",
    "\n",
    "We're going to use Python to grab the file directly from the Web using the 'requests' library.\n",
    "\n",
    "And finally, we're going to briefly introduce the Pandas library to peak at our output CSV.\n",
    "\n",
    "This file is also from the U.S. Bureau of Labor Statistics, the Local Area Unemployment program.\n",
    "\n",
    "The file layout is here\n",
    "https://download.bls.gov/pub/time.series/la/la.txt\n",
    "\n",
    "The key section:\n",
    "\n",
    "Field #/Data Element\tLength\t\tValue (Example)\t\t\n",
    "\n",
    "1. series_id\t\t  20\t\tLASBS060000000000003          \n",
    "\n",
    "2. year\t\t\t   4\t\t1976\n",
    "\n",
    "3. period\t\t   3\t\tM01\n",
    "\n",
    "4. value\t\t  12      \t9.4\n",
    "\t\t\t\t \n",
    "5. footnote_codes\t  10\t\tIt varies\n",
    "\t\t\t\t\n",
    "\n",
    "The series_id (LASBS060000000000003) can be broken out into:\n",
    "\n",
    "Code\t\t\t\t\tValue\n",
    "\n",
    "survey abbreviation\t=\t\tLA\n",
    "seasonal\t\t=\t\tS\n",
    "area_code\t\t=\t\tBS0600000000000\n",
    "measure_code\t\t=\t\t03\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#let's grab the file and save it locally\n",
    "import requests  # the lib that handles the url stuff\n",
    "\n",
    "r = requests.get(\"https://download.bls.gov/pub/time.series/la/la.data.0.CurrentU15-19\")  \n",
    "with open(r'/Users/robertgebeloff/Downloads/lau.txt', 'wb') as f: #change to your path\n",
    "    f.write(r.content)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1318605\n"
     ]
    }
   ],
   "source": [
    "# how many lines are in there? \n",
    "text_file = open(\"/Users/200545/Downloads/lau.txt\", \"r\") #change to your path\n",
    "lines = text_file.readlines()\n",
    "print len(lines)\n",
    "text_file.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#let's convert this into a csv file\n",
    "import csv\n",
    "#create a csv file\n",
    "mycsv = csv.writer(open('/Users/200545/Downloads/lau.csv', 'wb')) #change to your path\n",
    "#write out the headers -- note we're going to split the original series id into components areacode and measurecode\n",
    "mycsv.writerow(['seriesid', 'areacode','measurecode','year','period','value'])\n",
    "#loop through the text file skipping the first row, which has the original headers\n",
    "for line in lines[1:len(lines)]:\n",
    "                #for each row -- split it using the tabs, the first piece is the series id\n",
    "                 seriesid=line.split(\"\\t\")[0]\n",
    "                #within seriesid, characters 3 through 18 are the area id\n",
    "                 areacode=line.split(\"\\t\")[0][3:18]\n",
    "                #with the seriesid, characters 18 through 20 are the measurecode\n",
    "                 measurecode=line.split(\"\\t\")[0][18:20]\n",
    "                #year, period and value are tabs 1,2 and 3    \n",
    "                 year=line.split(\"\\t\")[1]\n",
    "                 period=line.split(\"\\t\")[2]\n",
    "                 value=line.split(\"\\t\")[3]\n",
    "                #write the row\n",
    "                 mycsv.writerow([seriesid, areacode,measurecode,year,period,value])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "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>seriesid</th>\n",
       "      <th>areacode</th>\n",
       "      <th>measurecode</th>\n",
       "      <th>year</th>\n",
       "      <th>period</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>LAUBS060000000000003</td>\n",
       "      <td>BS0600000000000</td>\n",
       "      <td>03</td>\n",
       "      <td>2015.0</td>\n",
       "      <td>M01</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>LAUBS060000000000003</td>\n",
       "      <td>BS0600000000000</td>\n",
       "      <td>03</td>\n",
       "      <td>2015.0</td>\n",
       "      <td>M02</td>\n",
       "      <td>6.7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>LAUBS060000000000003</td>\n",
       "      <td>BS0600000000000</td>\n",
       "      <td>03</td>\n",
       "      <td>2015.0</td>\n",
       "      <td>M03</td>\n",
       "      <td>6.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>LAUBS060000000000003</td>\n",
       "      <td>BS0600000000000</td>\n",
       "      <td>03</td>\n",
       "      <td>2015.0</td>\n",
       "      <td>M04</td>\n",
       "      <td>6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>LAUBS060000000000003</td>\n",
       "      <td>BS0600000000000</td>\n",
       "      <td>03</td>\n",
       "      <td>2015.0</td>\n",
       "      <td>M05</td>\n",
       "      <td>5.9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               seriesid         areacode measurecode    year period  value\n",
       "0  LAUBS060000000000003  BS0600000000000          03  2015.0    M01    7.0\n",
       "1  LAUBS060000000000003  BS0600000000000          03  2015.0    M02    6.7\n",
       "2  LAUBS060000000000003  BS0600000000000          03  2015.0    M03    6.6\n",
       "3  LAUBS060000000000003  BS0600000000000          03  2015.0    M04    6.0\n",
       "4  LAUBS060000000000003  BS0600000000000          03  2015.0    M05    5.9"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#lets open the csv with the Pandas library to see what it looks like\n",
    "import pandas as pd\n",
    "#this command opens the csv file and assigns datatypes to the columns, string 20, string 15, etc\n",
    "laus = pd.read_csv('/Users/200545/Downloads/lau.csv', dtype={'seriesid': 'S20', 'areacode':'S15', 'measurecode':'S2', 'period':'S3'})\n",
    "laus.head()\n",
    "#this is the Python equivalent of a spreadsheet or database table -- you'll learn more about working with these\n",
    "#in later classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}