{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 012_importing_datasets\n",
    "\n",
    "[Source](https://github.com/iArunava/Python-TheNoTheoryGuide/)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Required Imports\n",
    "import pandas as pd\n",
    "import sklearn as sk\n",
    "import sqlite3\n",
    "from pandas.io import sql"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Importing CSV files from local directory\n",
    "# NOTE: Make sure the Path you use contains the dataset named 'whereisthatdataset.csv'\n",
    "\n",
    "df1 = pd.read_csv ('./assets/whereisthatdataset.csv')               # Using relative path\n",
    "df2 = pd.read_csv ('/home/arunava/Datasets/whereisthatdataset.csv') # Using absolute path\n",
    "\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# If a dataset comes without headers then you need to pass `headers=None`\n",
    "# Note: This Dataset comes with headers, \n",
    "#       specifying `headers=None` leads python to treat the first row as part of the dataset\n",
    "\n",
    "df1 = pd.read_csv ('./assets/whereisthatdataset.csv', header=None)\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Specify header names while importing datasets with (or without) headers\n",
    "df1 = pd.read_csv ('./assets/whereisthatdataset.csv', header=None, names=['Where', 'on', 'earth', 'did', 'you', 'got', 'this', 'dataset', 'of', 'Pigeons', 'racing'])\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Importing file from URL\n",
    "df1 = pd.read_csv('https://raw.githubusercontent.com/iArunava/Python-TheNoTheoryGuide/master/assets/whereisthatdataset.csv')\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Reading Data from text file\n",
    "# NOTE: Use `sep` to specify how your data is seperated\n",
    "\n",
    "df1 = pd.read_table ('./assets/whereisthatdataset.txt', sep=',')\n",
    "df2 = pd.read_csv ('./assets/whereisthatdataset.txt', sep=',')\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read excel file\n",
    "# NOTE: you need 'xlrd' module to read .xls files\n",
    "\n",
    "df1 = pd.read_excel ('./assets/whereisthatdataset.xls', sheetname='whereisthatdataset', skiprows=1)\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read SAS file\n",
    "df1 = pd.read_sas ('./assets/whereisthatdataset.sas7bdat')\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read SQL Table\n",
    "\n",
    "conn  = sqlite3.connect ('./assets/whereisthatdataset.db')\n",
    "query = 'SELECT * FROM whereisthattable;'\n",
    "df1   = pd.read_sql(query, con=conn)\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read sample rows and columns\n",
    "# nrows:   Number of rows to select\n",
    "# usecols: list of cols to use (either all string or unicode)\n",
    "\n",
    "sdf1 = pd.read_csv ('./assets/whereisthatdataset.csv', nrows=4, usecols=[1, 5, 7])\n",
    "sdf2 = pd.read_csv ('./assets/whereisthatdataset.csv', nrows=4, usecols=['Breeder', 'Sex', 'Arrival'])\n",
    "sdf1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Skip rows while importing\n",
    "# NOTE: If you don't set header=None, pandas will treat the first row of all the rows to be considered as the header row\n",
    "\n",
    "df1 = pd.read_csv ('./assets/whereisthatdataset.csv', header=None, skiprows=5)\n",
    "df1.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Specify Missing Values\n",
    "# na_values: pass a list, which if present in the dataset will be considered as missing values\n",
    "\n",
    "df1 = pd.read_csv ('./assets/whereisthatdataset.csv', na_values=['NaN'])\n",
    "df1.head(3)"
   ]
  }
 ],
 "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}