diff --git a/notebooks/suresh_template.ipynb b/notebooks/suresh_template.ipynb new file mode 100644 index 0000000..509ff72 --- /dev/null +++ b/notebooks/suresh_template.ipynb @@ -0,0 +1,2907 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Title" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "2017-07-03 16:52:06 \n", + "\n", + "CPython 3.6.1\n", + "IPython 6.1.0\n", + "\n", + "pea 0.0.7\n", + "pandas 0.20.2\n", + "\n", + "compiler : MSC v.1900 64 bit (AMD64)\n", + "system : Windows\n", + "release : 7\n", + "machine : AMD64\n", + "processor : Intel64 Family 6 Model 42 Stepping 7, GenuineIntel\n", + "CPU cores : 8\n", + "interpreter: 64bit\n" + ] + } + ], + "source": [ + "import os\n", + "import pandas as pd\n", + "import settings\n", + "import etl\n", + "\n", + "%matplotlib inline\n", + "\n", + "%load_ext watermark\n", + "%watermark -d -t -v -m -p pea,pandas" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "data = etl.Data()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Available DataFrames\n", + "\n", + "the Data object has 3 DataFrames available movie, keyword, and genre. keyword and genre can be linked back to movie by the index column and the index of movie" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
colordirector_namenum_critic_for_reviewsdurationdirector_facebook_likesactor_3_facebook_likesactor_2_nameactor_1_facebook_likesgrossgenres...countrycontent_ratingbudgettitle_yearactor_2_facebook_likesimdb_scoreaspect_ratiomovie_facebook_likesnetprofitable
0ColorJames Cameron723.0178.00.0855.0Joel David Moore1000.0760505847.0Action|Adventure|Fantasy|Sci-Fi...USAPG-13237000000.02009-01-01936.07.91.7833000523505847.01
1ColorGore Verbinski302.0169.0563.01000.0Orlando Bloom40000.0309404152.0Action|Adventure|Fantasy...USAPG-13300000000.02007-01-015000.07.12.3509404152.01
2ColorSam Mendes602.0148.00.0161.0Rory Kinnear11000.0200074175.0Action|Adventure|Thriller...UKPG-13245000000.02015-01-01393.06.82.3585000-44925825.00
3ColorChristopher Nolan813.0164.022000.023000.0Christian Bale27000.0448130642.0Action|Thriller...USAPG-13250000000.02012-01-0123000.08.52.35164000198130642.01
4NaNDoug WalkerNaNNaN131.0NaNRob Walker131.0NaNDocumentary...NaNNaNNaNNaT12.07.1NaN0NaN0
5ColorAndrew Stanton462.0132.0475.0530.0Samantha Morton640.073058679.0Action|Adventure|Sci-Fi...USAPG-13263700000.02012-01-01632.06.62.3524000-190641321.00
6ColorSam Raimi392.0156.00.04000.0James Franco24000.0336530303.0Action|Adventure|Romance...USAPG-13258000000.02007-01-0111000.06.22.35078530303.01
7ColorNathan Greno324.0100.015.0284.0Donna Murphy799.0200807262.0Adventure|Animation|Comedy|Family|Fantasy|Musi......USAPG260000000.02010-01-01553.07.81.8529000-59192738.00
8ColorJoss Whedon635.0141.00.019000.0Robert Downey Jr.26000.0458991599.0Action|Adventure|Sci-Fi...USAPG-13250000000.02015-01-0121000.07.52.35118000208991599.01
9ColorDavid Yates375.0153.0282.010000.0Daniel Radcliffe25000.0301956980.0Adventure|Family|Fantasy|Mystery...UKPG250000000.02009-01-0111000.07.52.351000051956980.01
10ColorZack Snyder673.0183.00.02000.0Lauren Cohan15000.0330249062.0Action|Adventure|Sci-Fi...USAPG-13250000000.02016-01-014000.06.92.3519700080249062.01
11ColorBryan Singer434.0169.00.0903.0Marlon Brando18000.0200069408.0Action|Adventure|Sci-Fi...USAPG-13209000000.02006-01-0110000.06.12.350-8930592.00
12ColorMarc Forster403.0106.0395.0393.0Mathieu Amalric451.0168368427.0Action|Adventure...UKPG-13200000000.02008-01-01412.06.72.350-31631573.00
13ColorGore Verbinski313.0151.0563.01000.0Orlando Bloom40000.0423032628.0Action|Adventure|Fantasy...USAPG-13225000000.02006-01-015000.07.32.355000198032628.01
14ColorGore Verbinski450.0150.0563.01000.0Ruth Wilson40000.089289910.0Action|Adventure|Western...USAPG-13215000000.02013-01-012000.06.52.3548000-125710090.00
15ColorZack Snyder733.0143.00.0748.0Christopher Meloni15000.0291021565.0Action|Adventure|Fantasy|Sci-Fi...USAPG-13225000000.02013-01-013000.07.22.3511800066021565.01
16ColorAndrew Adamson258.0150.080.0201.0Pierfrancesco Favino22000.0141614023.0Action|Adventure|Family|Fantasy...USAPG225000000.02008-01-01216.06.62.350-83385977.00
17ColorJoss Whedon703.0173.00.019000.0Robert Downey Jr.26000.0623279547.0Action|Adventure|Sci-Fi...USAPG-13220000000.02012-01-0121000.08.11.85123000403279547.01
18ColorRob Marshall448.0136.0252.01000.0Sam Claflin40000.0241063875.0Action|Adventure|Fantasy...USAPG-13250000000.02011-01-0111000.06.72.3558000-8936125.00
19ColorBarry Sonnenfeld451.0106.0188.0718.0Michael Stuhlbarg10000.0179020854.0Action|Adventure|Comedy|Family|Fantasy|Sci-Fi...USAPG-13225000000.02012-01-01816.06.81.8540000-45979146.00
20ColorPeter Jackson422.0164.00.0773.0Adam Brown5000.0255108370.0Adventure|Fantasy...New ZealandPG-13250000000.02014-01-01972.07.52.35650005108370.01
21ColorMarc Webb599.0153.0464.0963.0Andrew Garfield15000.0262030663.0Action|Adventure|Fantasy...USAPG-13230000000.02012-01-0110000.07.02.355600032030663.01
22ColorRidley Scott343.0156.00.0738.0William Hurt891.0105219735.0Action|Adventure|Drama|History...USAPG-13200000000.02010-01-01882.06.72.3517000-94780265.00
23ColorPeter Jackson509.0186.00.0773.0Adam Brown5000.0258355354.0Adventure|Fantasy...USAPG-13225000000.02013-01-01972.07.92.358300033355354.01
24ColorChris Weitz251.0113.0129.01000.0Eva Green16000.070083519.0Adventure|Family|Fantasy...USAPG-13180000000.02007-01-016000.06.12.350-109916481.00
25ColorPeter Jackson446.0201.00.084.0Thomas Kretschmann6000.0218051260.0Action|Adventure|Drama|Romance...New ZealandPG-13207000000.02005-01-01919.07.22.35011051260.01
26ColorJames Cameron315.0194.00.0794.0Kate Winslet29000.0658672302.0Drama|Romance...USAPG-13200000000.01997-01-0114000.07.72.3526000458672302.01
27ColorAnthony Russo516.0147.094.011000.0Scarlett Johansson21000.0407197282.0Action|Adventure|Sci-Fi...USAPG-13250000000.02016-01-0119000.08.22.3572000157197282.01
28ColorPeter Berg377.0131.0532.0627.0Alexander Skarsgård14000.065173160.0Action|Adventure|Sci-Fi|Thriller...USAPG-13209000000.02012-01-0110000.05.92.3544000-143826840.00
29ColorColin Trevorrow644.0124.0365.01000.0Judy Greer3000.0652177271.0Action|Adventure|Sci-Fi|Thriller...USAPG-13150000000.02015-01-012000.07.02.00150000502177271.01
..................................................................
5013ColorEric Eason28.079.03.042.0Panchito Gómez93.0NaNDrama|Family...USANaN24000.02002-01-0146.07.01.7861NaN0
5014ColorUwe Boll58.080.0892.0492.0Katharine Isabelle986.0NaNAction|Crime|Thriller...CanadaRNaN2009-01-01918.06.32.350NaN0
5015Black and WhiteRichard Linklater61.0100.00.00.0Richard Linklater5.01227508.0Comedy|Drama...USAR23000.01991-01-010.07.11.3720001204508.01
5016ColorJoseph MazzellaNaN90.00.09.0Mikaal Bates313.0NaNCrime|Drama|Thriller...USANaN25000.02015-01-0125.04.8NaN33NaN0
5017ColorTravis Legge1.090.0138.0138.0Suzi Lorraine370.0NaNComedy|Romance...USANaN22000.02013-01-01184.03.31.78200NaN0
5018ColorAlex Kendrick5.0120.0589.04.0Lisa Arnold51.0NaNDrama...USANaN20000.02003-01-0149.06.91.85725NaN0
5019ColorMarcus Nispel43.091.0158.0265.0Brittany Curran630.0NaNHorror|Mystery|Thriller...USARNaN2015-01-01512.04.61.850NaN0
5020NaNBrandon LandersNaN143.08.08.0Alana Kaniewski720.0NaNDrama|Horror|Thriller...USANaN17350.02011-01-0119.03.0NaN33NaN0
5021ColorJay Duplass51.085.0157.010.0Katie Aselton830.0192467.0Comedy|Drama|Romance...USAR15000.02005-01-01224.06.6NaN297177467.01
5022Black and WhiteJim Chuchu6.060.00.04.0Olwenya Maina147.0NaNDrama...KenyaNaN15000.02014-01-0119.07.4NaN45NaN0
5023ColorDaryl Wein22.088.038.0211.0Heather Burns331.076382.0Romance...USANaN15000.02009-01-01212.06.22.3532461382.01
5024ColorJason Trost42.078.091.086.0Jason Trost407.0NaNSci-Fi|Thriller...USAUnrated20000.02011-01-0191.04.02.35835NaN0
5025ColorJohn Waters73.0108.00.0105.0Mink Stole462.0180483.0Comedy|Crime|Horror...USANC-1710000.01972-01-01143.06.11.370170483.01
5026ColorOlivier Assayas81.0110.0107.045.0Béatrice Dalle576.0136007.0Drama|Music|Romance...FranceR4500.02004-01-01133.06.92.35171131507.01
5027ColorJafar Panahi64.090.0397.00.0Nargess Mamizadeh5.0673780.0Drama...IranNot Rated10000.02000-01-010.07.51.85697663780.01
5028Black and WhiteIvan Kavanagh12.083.018.00.0Michael Parle10.0NaNHorror...IrelandNaN10000.02007-01-015.06.71.33105NaN0
5029ColorKiyoshi Kurosawa78.0111.062.06.0Anna Nakagawa89.094596.0Crime|Horror|Mystery|Thriller...JapanNaN1000000.01997-01-0113.07.41.85817-905404.00
5030ColorTadeo GarciaNaN84.05.012.0Michael Cortez21.0NaNDrama...USANaNNaN2004-01-0120.06.1NaN22NaN0
5031ColorThomas L. Phillips13.082.0120.084.0Joe Coffey785.0NaNComedy|Horror|Thriller...USANaN200000.02012-01-0198.05.416.00424NaN0
5032ColorAsh Baron-Cohen10.098.03.0152.0Stanley B. Herman789.0NaNCrime|Drama...USANaNNaN1995-01-01194.06.4NaN20NaN0
5033ColorShane Carruth143.077.0291.08.0David Sullivan291.0424760.0Drama|Sci-Fi|Thriller...USAPG-137000.02004-01-0145.07.01.8519000417760.01
5034ColorNeill Dela Llana35.080.00.00.0Edgar Tancangco0.070071.0Thriller...PhilippinesNot Rated7000.02005-01-010.06.3NaN7463071.01
5035ColorRobert Rodriguez56.081.00.06.0Peter Marquardt121.02040920.0Action|Crime|Drama|Romance|Thriller...USAR7000.01992-01-0120.06.91.3702033920.01
5036ColorAnthony ValloneNaN84.02.02.0John Considine45.0NaNCrime|Drama...USAPG-133250.02005-01-0144.07.8NaN4NaN0
5037ColorEdward Burns14.095.00.0133.0Caitlin FitzGerald296.04584.0Comedy|Drama...USANot Rated9000.02011-01-01205.06.4NaN413-4416.00
5038ColorScott Smith1.087.02.0318.0Daphne Zuniga637.0NaNComedy|Drama...CanadaNaNNaN2013-01-01470.07.7NaN84NaN0
5039ColorNaN43.043.0NaN319.0Valorie Curry841.0NaNCrime|Drama|Mystery|Thriller...USATV-14NaNNaT593.07.516.0032000NaN0
5040ColorBenjamin Roberds13.076.00.00.0Maxwell Moody0.0NaNDrama|Horror|Thriller...USANaN1400.02013-01-010.06.3NaN16NaN0
5041ColorDaniel Hsia14.0100.00.0489.0Daniel Henney946.010443.0Comedy|Drama|Romance...USAPG-13NaN2012-01-01719.06.32.35660NaN0
5042ColorJon Gunn43.090.016.016.0Brian Herzlinger86.085222.0Documentary...USAPG1100.02004-01-0123.06.61.8545684122.01
\n", + "

5043 rows × 30 columns

\n", + "
" + ], + "text/plain": [ + " color director_name num_critic_for_reviews duration \\\n", + "0 Color James Cameron 723.0 178.0 \n", + "1 Color Gore Verbinski 302.0 169.0 \n", + "2 Color Sam Mendes 602.0 148.0 \n", + "3 Color Christopher Nolan 813.0 164.0 \n", + "4 NaN Doug Walker NaN NaN \n", + "5 Color Andrew Stanton 462.0 132.0 \n", + "6 Color Sam Raimi 392.0 156.0 \n", + "7 Color Nathan Greno 324.0 100.0 \n", + "8 Color Joss Whedon 635.0 141.0 \n", + "9 Color David Yates 375.0 153.0 \n", + "10 Color Zack Snyder 673.0 183.0 \n", + "11 Color Bryan Singer 434.0 169.0 \n", + "12 Color Marc Forster 403.0 106.0 \n", + "13 Color Gore Verbinski 313.0 151.0 \n", + "14 Color Gore Verbinski 450.0 150.0 \n", + "15 Color Zack Snyder 733.0 143.0 \n", + "16 Color Andrew Adamson 258.0 150.0 \n", + "17 Color Joss Whedon 703.0 173.0 \n", + "18 Color Rob Marshall 448.0 136.0 \n", + "19 Color Barry Sonnenfeld 451.0 106.0 \n", + "20 Color Peter Jackson 422.0 164.0 \n", + "21 Color Marc Webb 599.0 153.0 \n", + "22 Color Ridley Scott 343.0 156.0 \n", + "23 Color Peter Jackson 509.0 186.0 \n", + "24 Color Chris Weitz 251.0 113.0 \n", + "25 Color Peter Jackson 446.0 201.0 \n", + "26 Color James Cameron 315.0 194.0 \n", + "27 Color Anthony Russo 516.0 147.0 \n", + "28 Color Peter Berg 377.0 131.0 \n", + "29 Color Colin Trevorrow 644.0 124.0 \n", + "... ... ... ... ... \n", + "5013 Color Eric Eason 28.0 79.0 \n", + "5014 Color Uwe Boll 58.0 80.0 \n", + "5015 Black and White Richard Linklater 61.0 100.0 \n", + "5016 Color Joseph Mazzella NaN 90.0 \n", + "5017 Color Travis Legge 1.0 90.0 \n", + "5018 Color Alex Kendrick 5.0 120.0 \n", + "5019 Color Marcus Nispel 43.0 91.0 \n", + "5020 NaN Brandon Landers NaN 143.0 \n", + "5021 Color Jay Duplass 51.0 85.0 \n", + "5022 Black and White Jim Chuchu 6.0 60.0 \n", + "5023 Color Daryl Wein 22.0 88.0 \n", + "5024 Color Jason Trost 42.0 78.0 \n", + "5025 Color John Waters 73.0 108.0 \n", + "5026 Color Olivier Assayas 81.0 110.0 \n", + "5027 Color Jafar Panahi 64.0 90.0 \n", + "5028 Black and White Ivan Kavanagh 12.0 83.0 \n", + "5029 Color Kiyoshi Kurosawa 78.0 111.0 \n", + "5030 Color Tadeo Garcia NaN 84.0 \n", + "5031 Color Thomas L. Phillips 13.0 82.0 \n", + "5032 Color Ash Baron-Cohen 10.0 98.0 \n", + "5033 Color Shane Carruth 143.0 77.0 \n", + "5034 Color Neill Dela Llana 35.0 80.0 \n", + "5035 Color Robert Rodriguez 56.0 81.0 \n", + "5036 Color Anthony Vallone NaN 84.0 \n", + "5037 Color Edward Burns 14.0 95.0 \n", + "5038 Color Scott Smith 1.0 87.0 \n", + "5039 Color NaN 43.0 43.0 \n", + "5040 Color Benjamin Roberds 13.0 76.0 \n", + "5041 Color Daniel Hsia 14.0 100.0 \n", + "5042 Color Jon Gunn 43.0 90.0 \n", + "\n", + " director_facebook_likes actor_3_facebook_likes actor_2_name \\\n", + "0 0.0 855.0 Joel David Moore \n", + "1 563.0 1000.0 Orlando Bloom \n", + "2 0.0 161.0 Rory Kinnear \n", + "3 22000.0 23000.0 Christian Bale \n", + "4 131.0 NaN Rob Walker \n", + "5 475.0 530.0 Samantha Morton \n", + "6 0.0 4000.0 James Franco \n", + "7 15.0 284.0 Donna Murphy \n", + "8 0.0 19000.0 Robert Downey Jr. \n", + "9 282.0 10000.0 Daniel Radcliffe \n", + "10 0.0 2000.0 Lauren Cohan \n", + "11 0.0 903.0 Marlon Brando \n", + "12 395.0 393.0 Mathieu Amalric \n", + "13 563.0 1000.0 Orlando Bloom \n", + "14 563.0 1000.0 Ruth Wilson \n", + "15 0.0 748.0 Christopher Meloni \n", + "16 80.0 201.0 Pierfrancesco Favino \n", + "17 0.0 19000.0 Robert Downey Jr. \n", + "18 252.0 1000.0 Sam Claflin \n", + "19 188.0 718.0 Michael Stuhlbarg \n", + "20 0.0 773.0 Adam Brown \n", + "21 464.0 963.0 Andrew Garfield \n", + "22 0.0 738.0 William Hurt \n", + "23 0.0 773.0 Adam Brown \n", + "24 129.0 1000.0 Eva Green \n", + "25 0.0 84.0 Thomas Kretschmann \n", + "26 0.0 794.0 Kate Winslet \n", + "27 94.0 11000.0 Scarlett Johansson \n", + "28 532.0 627.0 Alexander Skarsgård \n", + "29 365.0 1000.0 Judy Greer \n", + "... ... ... ... \n", + "5013 3.0 42.0 Panchito Gómez \n", + "5014 892.0 492.0 Katharine Isabelle \n", + "5015 0.0 0.0 Richard Linklater \n", + "5016 0.0 9.0 Mikaal Bates \n", + "5017 138.0 138.0 Suzi Lorraine \n", + "5018 589.0 4.0 Lisa Arnold \n", + "5019 158.0 265.0 Brittany Curran \n", + "5020 8.0 8.0 Alana Kaniewski \n", + "5021 157.0 10.0 Katie Aselton \n", + "5022 0.0 4.0 Olwenya Maina \n", + "5023 38.0 211.0 Heather Burns \n", + "5024 91.0 86.0 Jason Trost \n", + "5025 0.0 105.0 Mink Stole \n", + "5026 107.0 45.0 Béatrice Dalle \n", + "5027 397.0 0.0 Nargess Mamizadeh \n", + "5028 18.0 0.0 Michael Parle \n", + "5029 62.0 6.0 Anna Nakagawa \n", + "5030 5.0 12.0 Michael Cortez \n", + "5031 120.0 84.0 Joe Coffey \n", + "5032 3.0 152.0 Stanley B. Herman \n", + "5033 291.0 8.0 David Sullivan \n", + "5034 0.0 0.0 Edgar Tancangco \n", + "5035 0.0 6.0 Peter Marquardt \n", + "5036 2.0 2.0 John Considine \n", + "5037 0.0 133.0 Caitlin FitzGerald \n", + "5038 2.0 318.0 Daphne Zuniga \n", + "5039 NaN 319.0 Valorie Curry \n", + "5040 0.0 0.0 Maxwell Moody \n", + "5041 0.0 489.0 Daniel Henney \n", + "5042 16.0 16.0 Brian Herzlinger \n", + "\n", + " actor_1_facebook_likes gross \\\n", + "0 1000.0 760505847.0 \n", + "1 40000.0 309404152.0 \n", + "2 11000.0 200074175.0 \n", + "3 27000.0 448130642.0 \n", + "4 131.0 NaN \n", + "5 640.0 73058679.0 \n", + "6 24000.0 336530303.0 \n", + "7 799.0 200807262.0 \n", + "8 26000.0 458991599.0 \n", + "9 25000.0 301956980.0 \n", + "10 15000.0 330249062.0 \n", + "11 18000.0 200069408.0 \n", + "12 451.0 168368427.0 \n", + "13 40000.0 423032628.0 \n", + "14 40000.0 89289910.0 \n", + "15 15000.0 291021565.0 \n", + "16 22000.0 141614023.0 \n", + "17 26000.0 623279547.0 \n", + "18 40000.0 241063875.0 \n", + "19 10000.0 179020854.0 \n", + "20 5000.0 255108370.0 \n", + "21 15000.0 262030663.0 \n", + "22 891.0 105219735.0 \n", + "23 5000.0 258355354.0 \n", + "24 16000.0 70083519.0 \n", + "25 6000.0 218051260.0 \n", + "26 29000.0 658672302.0 \n", + "27 21000.0 407197282.0 \n", + "28 14000.0 65173160.0 \n", + "29 3000.0 652177271.0 \n", + "... ... ... \n", + "5013 93.0 NaN \n", + "5014 986.0 NaN \n", + "5015 5.0 1227508.0 \n", + "5016 313.0 NaN \n", + "5017 370.0 NaN \n", + "5018 51.0 NaN \n", + "5019 630.0 NaN \n", + "5020 720.0 NaN \n", + "5021 830.0 192467.0 \n", + "5022 147.0 NaN \n", + "5023 331.0 76382.0 \n", + "5024 407.0 NaN \n", + "5025 462.0 180483.0 \n", + "5026 576.0 136007.0 \n", + "5027 5.0 673780.0 \n", + "5028 10.0 NaN \n", + "5029 89.0 94596.0 \n", + "5030 21.0 NaN \n", + "5031 785.0 NaN \n", + "5032 789.0 NaN \n", + "5033 291.0 424760.0 \n", + "5034 0.0 70071.0 \n", + "5035 121.0 2040920.0 \n", + "5036 45.0 NaN \n", + "5037 296.0 4584.0 \n", + "5038 637.0 NaN \n", + "5039 841.0 NaN \n", + "5040 0.0 NaN \n", + "5041 946.0 10443.0 \n", + "5042 86.0 85222.0 \n", + "\n", + " genres ... \\\n", + "0 Action|Adventure|Fantasy|Sci-Fi ... \n", + "1 Action|Adventure|Fantasy ... \n", + "2 Action|Adventure|Thriller ... \n", + "3 Action|Thriller ... \n", + "4 Documentary ... \n", + "5 Action|Adventure|Sci-Fi ... \n", + "6 Action|Adventure|Romance ... \n", + "7 Adventure|Animation|Comedy|Family|Fantasy|Musi... ... \n", + "8 Action|Adventure|Sci-Fi ... \n", + "9 Adventure|Family|Fantasy|Mystery ... \n", + "10 Action|Adventure|Sci-Fi ... \n", + "11 Action|Adventure|Sci-Fi ... \n", + "12 Action|Adventure ... \n", + "13 Action|Adventure|Fantasy ... \n", + "14 Action|Adventure|Western ... \n", + "15 Action|Adventure|Fantasy|Sci-Fi ... \n", + "16 Action|Adventure|Family|Fantasy ... \n", + "17 Action|Adventure|Sci-Fi ... \n", + "18 Action|Adventure|Fantasy ... \n", + "19 Action|Adventure|Comedy|Family|Fantasy|Sci-Fi ... \n", + "20 Adventure|Fantasy ... \n", + "21 Action|Adventure|Fantasy ... \n", + "22 Action|Adventure|Drama|History ... \n", + "23 Adventure|Fantasy ... \n", + "24 Adventure|Family|Fantasy ... \n", + "25 Action|Adventure|Drama|Romance ... \n", + "26 Drama|Romance ... \n", + "27 Action|Adventure|Sci-Fi ... \n", + "28 Action|Adventure|Sci-Fi|Thriller ... \n", + "29 Action|Adventure|Sci-Fi|Thriller ... \n", + "... ... ... \n", + "5013 Drama|Family ... \n", + "5014 Action|Crime|Thriller ... \n", + "5015 Comedy|Drama ... \n", + "5016 Crime|Drama|Thriller ... \n", + "5017 Comedy|Romance ... \n", + "5018 Drama ... \n", + "5019 Horror|Mystery|Thriller ... \n", + "5020 Drama|Horror|Thriller ... \n", + "5021 Comedy|Drama|Romance ... \n", + "5022 Drama ... \n", + "5023 Romance ... \n", + "5024 Sci-Fi|Thriller ... \n", + "5025 Comedy|Crime|Horror ... \n", + "5026 Drama|Music|Romance ... \n", + "5027 Drama ... \n", + "5028 Horror ... \n", + "5029 Crime|Horror|Mystery|Thriller ... \n", + "5030 Drama ... \n", + "5031 Comedy|Horror|Thriller ... \n", + "5032 Crime|Drama ... \n", + "5033 Drama|Sci-Fi|Thriller ... \n", + "5034 Thriller ... \n", + "5035 Action|Crime|Drama|Romance|Thriller ... \n", + "5036 Crime|Drama ... \n", + "5037 Comedy|Drama ... \n", + "5038 Comedy|Drama ... \n", + "5039 Crime|Drama|Mystery|Thriller ... \n", + "5040 Drama|Horror|Thriller ... \n", + "5041 Comedy|Drama|Romance ... \n", + "5042 Documentary ... \n", + "\n", + " country content_rating budget title_year \\\n", + "0 USA PG-13 237000000.0 2009-01-01 \n", + "1 USA PG-13 300000000.0 2007-01-01 \n", + "2 UK PG-13 245000000.0 2015-01-01 \n", + "3 USA PG-13 250000000.0 2012-01-01 \n", + "4 NaN NaN NaN NaT \n", + "5 USA PG-13 263700000.0 2012-01-01 \n", + "6 USA PG-13 258000000.0 2007-01-01 \n", + "7 USA PG 260000000.0 2010-01-01 \n", + "8 USA PG-13 250000000.0 2015-01-01 \n", + "9 UK PG 250000000.0 2009-01-01 \n", + "10 USA PG-13 250000000.0 2016-01-01 \n", + "11 USA PG-13 209000000.0 2006-01-01 \n", + "12 UK PG-13 200000000.0 2008-01-01 \n", + "13 USA PG-13 225000000.0 2006-01-01 \n", + "14 USA PG-13 215000000.0 2013-01-01 \n", + "15 USA PG-13 225000000.0 2013-01-01 \n", + "16 USA PG 225000000.0 2008-01-01 \n", + "17 USA PG-13 220000000.0 2012-01-01 \n", + "18 USA PG-13 250000000.0 2011-01-01 \n", + "19 USA PG-13 225000000.0 2012-01-01 \n", + "20 New Zealand PG-13 250000000.0 2014-01-01 \n", + "21 USA PG-13 230000000.0 2012-01-01 \n", + "22 USA PG-13 200000000.0 2010-01-01 \n", + "23 USA PG-13 225000000.0 2013-01-01 \n", + "24 USA PG-13 180000000.0 2007-01-01 \n", + "25 New Zealand PG-13 207000000.0 2005-01-01 \n", + "26 USA PG-13 200000000.0 1997-01-01 \n", + "27 USA PG-13 250000000.0 2016-01-01 \n", + "28 USA PG-13 209000000.0 2012-01-01 \n", + "29 USA PG-13 150000000.0 2015-01-01 \n", + "... ... ... ... ... \n", + "5013 USA NaN 24000.0 2002-01-01 \n", + "5014 Canada R NaN 2009-01-01 \n", + "5015 USA R 23000.0 1991-01-01 \n", + "5016 USA NaN 25000.0 2015-01-01 \n", + "5017 USA NaN 22000.0 2013-01-01 \n", + "5018 USA NaN 20000.0 2003-01-01 \n", + "5019 USA R NaN 2015-01-01 \n", + "5020 USA NaN 17350.0 2011-01-01 \n", + "5021 USA R 15000.0 2005-01-01 \n", + "5022 Kenya NaN 15000.0 2014-01-01 \n", + "5023 USA NaN 15000.0 2009-01-01 \n", + "5024 USA Unrated 20000.0 2011-01-01 \n", + "5025 USA NC-17 10000.0 1972-01-01 \n", + "5026 France R 4500.0 2004-01-01 \n", + "5027 Iran Not Rated 10000.0 2000-01-01 \n", + "5028 Ireland NaN 10000.0 2007-01-01 \n", + "5029 Japan NaN 1000000.0 1997-01-01 \n", + "5030 USA NaN NaN 2004-01-01 \n", + "5031 USA NaN 200000.0 2012-01-01 \n", + "5032 USA NaN NaN 1995-01-01 \n", + "5033 USA PG-13 7000.0 2004-01-01 \n", + "5034 Philippines Not Rated 7000.0 2005-01-01 \n", + "5035 USA R 7000.0 1992-01-01 \n", + "5036 USA PG-13 3250.0 2005-01-01 \n", + "5037 USA Not Rated 9000.0 2011-01-01 \n", + "5038 Canada NaN NaN 2013-01-01 \n", + "5039 USA TV-14 NaN NaT \n", + "5040 USA NaN 1400.0 2013-01-01 \n", + "5041 USA PG-13 NaN 2012-01-01 \n", + "5042 USA PG 1100.0 2004-01-01 \n", + "\n", + " actor_2_facebook_likes imdb_score aspect_ratio movie_facebook_likes \\\n", + "0 936.0 7.9 1.78 33000 \n", + "1 5000.0 7.1 2.35 0 \n", + "2 393.0 6.8 2.35 85000 \n", + "3 23000.0 8.5 2.35 164000 \n", + "4 12.0 7.1 NaN 0 \n", + "5 632.0 6.6 2.35 24000 \n", + "6 11000.0 6.2 2.35 0 \n", + "7 553.0 7.8 1.85 29000 \n", + "8 21000.0 7.5 2.35 118000 \n", + "9 11000.0 7.5 2.35 10000 \n", + "10 4000.0 6.9 2.35 197000 \n", + "11 10000.0 6.1 2.35 0 \n", + "12 412.0 6.7 2.35 0 \n", + "13 5000.0 7.3 2.35 5000 \n", + "14 2000.0 6.5 2.35 48000 \n", + "15 3000.0 7.2 2.35 118000 \n", + "16 216.0 6.6 2.35 0 \n", + "17 21000.0 8.1 1.85 123000 \n", + "18 11000.0 6.7 2.35 58000 \n", + "19 816.0 6.8 1.85 40000 \n", + "20 972.0 7.5 2.35 65000 \n", + "21 10000.0 7.0 2.35 56000 \n", + "22 882.0 6.7 2.35 17000 \n", + "23 972.0 7.9 2.35 83000 \n", + "24 6000.0 6.1 2.35 0 \n", + "25 919.0 7.2 2.35 0 \n", + "26 14000.0 7.7 2.35 26000 \n", + "27 19000.0 8.2 2.35 72000 \n", + "28 10000.0 5.9 2.35 44000 \n", + "29 2000.0 7.0 2.00 150000 \n", + "... ... ... ... ... \n", + "5013 46.0 7.0 1.78 61 \n", + "5014 918.0 6.3 2.35 0 \n", + "5015 0.0 7.1 1.37 2000 \n", + "5016 25.0 4.8 NaN 33 \n", + "5017 184.0 3.3 1.78 200 \n", + "5018 49.0 6.9 1.85 725 \n", + "5019 512.0 4.6 1.85 0 \n", + "5020 19.0 3.0 NaN 33 \n", + "5021 224.0 6.6 NaN 297 \n", + "5022 19.0 7.4 NaN 45 \n", + "5023 212.0 6.2 2.35 324 \n", + "5024 91.0 4.0 2.35 835 \n", + "5025 143.0 6.1 1.37 0 \n", + "5026 133.0 6.9 2.35 171 \n", + "5027 0.0 7.5 1.85 697 \n", + "5028 5.0 6.7 1.33 105 \n", + "5029 13.0 7.4 1.85 817 \n", + "5030 20.0 6.1 NaN 22 \n", + "5031 98.0 5.4 16.00 424 \n", + "5032 194.0 6.4 NaN 20 \n", + "5033 45.0 7.0 1.85 19000 \n", + "5034 0.0 6.3 NaN 74 \n", + "5035 20.0 6.9 1.37 0 \n", + "5036 44.0 7.8 NaN 4 \n", + "5037 205.0 6.4 NaN 413 \n", + "5038 470.0 7.7 NaN 84 \n", + "5039 593.0 7.5 16.00 32000 \n", + "5040 0.0 6.3 NaN 16 \n", + "5041 719.0 6.3 2.35 660 \n", + "5042 23.0 6.6 1.85 456 \n", + "\n", + " net profitable \n", + "0 523505847.0 1 \n", + "1 9404152.0 1 \n", + "2 -44925825.0 0 \n", + "3 198130642.0 1 \n", + "4 NaN 0 \n", + "5 -190641321.0 0 \n", + "6 78530303.0 1 \n", + "7 -59192738.0 0 \n", + "8 208991599.0 1 \n", + "9 51956980.0 1 \n", + "10 80249062.0 1 \n", + "11 -8930592.0 0 \n", + "12 -31631573.0 0 \n", + "13 198032628.0 1 \n", + "14 -125710090.0 0 \n", + "15 66021565.0 1 \n", + "16 -83385977.0 0 \n", + "17 403279547.0 1 \n", + "18 -8936125.0 0 \n", + "19 -45979146.0 0 \n", + "20 5108370.0 1 \n", + "21 32030663.0 1 \n", + "22 -94780265.0 0 \n", + "23 33355354.0 1 \n", + "24 -109916481.0 0 \n", + "25 11051260.0 1 \n", + "26 458672302.0 1 \n", + "27 157197282.0 1 \n", + "28 -143826840.0 0 \n", + "29 502177271.0 1 \n", + "... ... ... \n", + "5013 NaN 0 \n", + "5014 NaN 0 \n", + "5015 1204508.0 1 \n", + "5016 NaN 0 \n", + "5017 NaN 0 \n", + "5018 NaN 0 \n", + "5019 NaN 0 \n", + "5020 NaN 0 \n", + "5021 177467.0 1 \n", + "5022 NaN 0 \n", + "5023 61382.0 1 \n", + "5024 NaN 0 \n", + "5025 170483.0 1 \n", + "5026 131507.0 1 \n", + "5027 663780.0 1 \n", + "5028 NaN 0 \n", + "5029 -905404.0 0 \n", + "5030 NaN 0 \n", + "5031 NaN 0 \n", + "5032 NaN 0 \n", + "5033 417760.0 1 \n", + "5034 63071.0 1 \n", + "5035 2033920.0 1 \n", + "5036 NaN 0 \n", + "5037 -4416.0 0 \n", + "5038 NaN 0 \n", + "5039 NaN 0 \n", + "5040 NaN 0 \n", + "5041 NaN 0 \n", + "5042 84122.0 1 \n", + "\n", + "[5043 rows x 30 columns]" + ] + }, + "execution_count": 4, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "data.movie" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
indexplot_keywords
00avatar
10future
20marine
30native
40paraplegic
51goddess
61marriage ceremony
71marriage proposal
81pirate
91singapore
102bomb
112espionage
122sequel
132spy
142terrorist
153deception
163imprisonment
173lawlessness
183police officer
193terrorist plot
204
215alien
225american civil war
235male nipple
245mars
255princess
266sandman
276spider man
286symbiote
296venom
.........
242205034security guard
242215034squatter
242225035assassin
242235035death
242245035guitar
242255035gun
242265035mariachi
242275036jewell
242285036mongol
242295036nostradamus
242305036stepnicka
242315036vallone
242325037written and directed by cast member
242335038fraud
242345038postal worker
242355038prison
242365038theft
242375038trial
242385039cult
242395039fbi
242405039hideout
242415039prison escape
242425039serial killer
242435040
242445041
242455042actress name in title
242465042crush
242475042date
242485042four word title
242495042video camera
\n", + "

24250 rows × 2 columns

\n", + "
" + ], + "text/plain": [ + " index plot_keywords\n", + "0 0 avatar\n", + "1 0 future\n", + "2 0 marine\n", + "3 0 native\n", + "4 0 paraplegic\n", + "5 1 goddess\n", + "6 1 marriage ceremony\n", + "7 1 marriage proposal\n", + "8 1 pirate\n", + "9 1 singapore\n", + "10 2 bomb\n", + "11 2 espionage\n", + "12 2 sequel\n", + "13 2 spy\n", + "14 2 terrorist\n", + "15 3 deception\n", + "16 3 imprisonment\n", + "17 3 lawlessness\n", + "18 3 police officer\n", + "19 3 terrorist plot\n", + "20 4 \n", + "21 5 alien\n", + "22 5 american civil war\n", + "23 5 male nipple\n", + "24 5 mars\n", + "25 5 princess\n", + "26 6 sandman\n", + "27 6 spider man\n", + "28 6 symbiote\n", + "29 6 venom\n", + "... ... ...\n", + "24220 5034 security guard\n", + "24221 5034 squatter\n", + "24222 5035 assassin\n", + "24223 5035 death\n", + "24224 5035 guitar\n", + "24225 5035 gun\n", + "24226 5035 mariachi\n", + "24227 5036 jewell\n", + "24228 5036 mongol\n", + "24229 5036 nostradamus\n", + "24230 5036 stepnicka\n", + "24231 5036 vallone\n", + "24232 5037 written and directed by cast member\n", + "24233 5038 fraud\n", + "24234 5038 postal worker\n", + "24235 5038 prison\n", + "24236 5038 theft\n", + "24237 5038 trial\n", + "24238 5039 cult\n", + "24239 5039 fbi\n", + "24240 5039 hideout\n", + "24241 5039 prison escape\n", + "24242 5039 serial killer\n", + "24243 5040 \n", + "24244 5041 \n", + "24245 5042 actress name in title\n", + "24246 5042 crush\n", + "24247 5042 date\n", + "24248 5042 four word title\n", + "24249 5042 video camera\n", + "\n", + "[24250 rows x 2 columns]" + ] + }, + "execution_count": 5, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "data.keyword" + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "metadata": {}, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
indexgenres
00Action
10Adventure
20Fantasy
30Sci-Fi
41Action
51Adventure
61Fantasy
72Action
82Adventure
92Thriller
103Action
113Thriller
124Documentary
135Action
145Adventure
155Sci-Fi
166Action
176Adventure
186Romance
197Adventure
207Animation
217Comedy
227Family
237Fantasy
247Musical
257Romance
268Action
278Adventure
288Sci-Fi
299Adventure
.........
144745031Horror
144755031Thriller
144765032Crime
144775032Drama
144785033Drama
144795033Sci-Fi
144805033Thriller
144815034Thriller
144825035Action
144835035Crime
144845035Drama
144855035Romance
144865035Thriller
144875036Crime
144885036Drama
144895037Comedy
144905037Drama
144915038Comedy
144925038Drama
144935039Crime
144945039Drama
144955039Mystery
144965039Thriller
144975040Drama
144985040Horror
144995040Thriller
145005041Comedy
145015041Drama
145025041Romance
145035042Documentary
\n", + "

14504 rows × 2 columns

\n", + "
" + ], + "text/plain": [ + " index genres\n", + "0 0 Action\n", + "1 0 Adventure\n", + "2 0 Fantasy\n", + "3 0 Sci-Fi\n", + "4 1 Action\n", + "5 1 Adventure\n", + "6 1 Fantasy\n", + "7 2 Action\n", + "8 2 Adventure\n", + "9 2 Thriller\n", + "10 3 Action\n", + "11 3 Thriller\n", + "12 4 Documentary\n", + "13 5 Action\n", + "14 5 Adventure\n", + "15 5 Sci-Fi\n", + "16 6 Action\n", + "17 6 Adventure\n", + "18 6 Romance\n", + "19 7 Adventure\n", + "20 7 Animation\n", + "21 7 Comedy\n", + "22 7 Family\n", + "23 7 Fantasy\n", + "24 7 Musical\n", + "25 7 Romance\n", + "26 8 Action\n", + "27 8 Adventure\n", + "28 8 Sci-Fi\n", + "29 9 Adventure\n", + "... ... ...\n", + "14474 5031 Horror\n", + "14475 5031 Thriller\n", + "14476 5032 Crime\n", + "14477 5032 Drama\n", + "14478 5033 Drama\n", + "14479 5033 Sci-Fi\n", + "14480 5033 Thriller\n", + "14481 5034 Thriller\n", + "14482 5035 Action\n", + "14483 5035 Crime\n", + "14484 5035 Drama\n", + "14485 5035 Romance\n", + "14486 5035 Thriller\n", + "14487 5036 Crime\n", + "14488 5036 Drama\n", + "14489 5037 Comedy\n", + "14490 5037 Drama\n", + "14491 5038 Comedy\n", + "14492 5038 Drama\n", + "14493 5039 Crime\n", + "14494 5039 Drama\n", + "14495 5039 Mystery\n", + "14496 5039 Thriller\n", + "14497 5040 Drama\n", + "14498 5040 Horror\n", + "14499 5040 Thriller\n", + "14500 5041 Comedy\n", + "14501 5041 Drama\n", + "14502 5041 Romance\n", + "14503 5042 Documentary\n", + "\n", + "[14504 rows x 2 columns]" + ] + }, + "execution_count": 6, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "data.genre" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python [default]", + "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": "3.6.1" + } + }, + "nbformat": 4, + "nbformat_minor": 2 +}