REEL RATINGS

Contributors:

• Andrew Polk
• Ari Gilmore
• Brandon Tran
• Danielle Robinson
• Ivy Tran
• Jake Hickey
• Victoria Sneddon (Project Manager)

Abstract

We want to be able to predict the rating of a movie before it is released. We want to be able to input a movie or upcoming movie title and output the predicted movie rating.

Table of Contents

• Flowchart
• Aspects
• Datasources
• Languages
• Blockers
• Setting Up Virtual Environment
• Running the Code
• Structure

Timeline

Fall 2017

• Week 1-3
• Week 4-6
• Week 7-10

Winter 2018

Project Flowchart

Aspects

• Rotten Tomatoes API (Kaggle, IMDB) -previous movies of that genre and their rating
• Data Algorithms (weighting actors vs. directors vs. budget…)
• Query data about movie title or movie id to get parameters
• Query previous ratings of rating parameters
• Create a method of storing data and accessing it from different parts of the program
• Some sort of statistical analysis on ratings (Data algorithms)
• Create an interface or web app to take input and display results

• Paramaters available to use:
  • Average movie rating
  • Actors
  • Director
  • Producers
  • Budget
  • Studio
  • Writers
  • Genre
  • Cinematographer
  • Music
  • etc..

Data Sources

• IMDb
• TheMovieDB

Languages

• Python - scripting
• HTML - website
• Java - AWS API

Blockers

• TheMovieDB has a 40 query per 10 second limit per api key.
• AWS requires paid subscription
• IMDb data must be refreshed manually

Creating a virtual environment

• First time run:

  • sudo apt install virtualenv (if needed)

  • virtualenv peerenv

• Starting virtualenv

  • source peerenv/bin/activate

  • pip install -r requirements.txt

• Closing virtualenv

  • deactivate

Running the code

• Make sure you have:

  • api_keys.txt
  • IMDBRatings.tsv
  • TMDBRatings.tsv

• Running in terminal

  • Start virtual environment

  • cd movie_rating_prediction/src/

  • python3 dataCall_example.py

  • Enter Movie Name

• Running in Idle

  • Start virtual environment

  • cd movie_rating_prediction/src/

  • idle3

  • Open dataCall_example.py
  • Run Module +Enter Movie Name

Project Timeline

Fall 2017

Week 1-3:

• Create an outline for the project
• Split into smaller groups (Pair up) to work on individual sections of the project

• Week 1-3 Update:

  • Brandon/Danielle:

    • Made an account for Amazon Web Services (AWS)
    • Downloaded AWS SDK for Java
    • Downloaded legacy Java 6 for Eclipse
    • Downloaded JVM 1.8 for Eclipse
    • Configured AWS Toolkit for Eclipse, and samples successfully
    • https://569112368633.signin.aws.amazon.com/console log in

  • Ari/Ivy/Jake:

    • Will weight the ratings of directors previous films based on date.
    • Take the sample mean of the ratings of the director’s movies and that will be the prediction
    • Award data? (how many oscars won) (later)
    • Age of the director (later)
    • Regression chart for when other factors come in

  • Andrew/Victoria:

    • Created a simple script in python, using libraries: json, requests, datetime
    • The script accesses themoviedb and get the works and ratings for a director parsed into json
    • Working on ways to pass the information from the script to the stats team
    • Still need to get director age, awards and any other data the stats team wants
    • Get data from rottenTomatoes/IMDB ect...?

Week 4-6:

• Start putting together the different portions that the smaller groups have been working on

• Week 4-6 Update:

  • Brandon/Danielle:

    • Downloaded IMDb's datasets using AWS S3 services
    • Gave Andrew the data to merge

  • Ari/Ivy/Jake:

    • Considering date of directors' previous movies as a factor in rating
    • Using number of ratings per movie rating to calculate mean value of directors' previous films

  • Andrew/Victoria:

    • Created script to access IMBDb's title names, ratings, etc.
    • Created script to merge ratings and title info tsv files
    • Will utilize both Movie DB and IMDb as foundation of project's data

Week 7-10:

• Finish merging different sections of the project

• Week 7-10 Update:

  • All:

    • Wrote data regression algorithm to weigh directors' previous films
    • Wrote scripting programs to input movie ID number and output parameters

Winter 2018

Goals:

• Finish scripting programs to fetch parameters from movie names
• Focus on documentation, here and on Inertia7

Structure

├── LICENSE
├── README.md          <- The top-level README for developers using this project.
│
├── data	<- We cant save the data online so this is empty
│   ├── external       <- Data from third party sources.
│   ├── interim        <- Intermediate data that has been transformed.
│   ├── processed      <- The final, canonical data sets for modeling.
│   └── raw            <- The original, immutable data dump.
│
├── references         <- Data dictionaries, manuals, and all other explanatory materials.
│
├── reports            <- Generated analysis as HTML, PDF, LaTeX, etc.
│   └── figures        <- Generated graphics and figures to be used in reporting
│
├── requirements.txt   <- The requirements file for reproducing the analysis environment, e.g.
│                         generated with `pip freeze > requirements.txt`
│
└── src                <- Source code for use in this project.
    ├── __init__.py    <- Makes src a Python module
    ├── dataCall_example.py    <- Whole script run from here
    │
    ├── api_keys.txt 	<- stores api keys to be used in query (Not online)
    ├── IMDBRatings.tsv <- stores the ratings sorted by IMDB ids
    ├── TMDBRatings.tsv <- stores the ratings sorted by TMDB ids
    │
    ├── data           <- Scripts to download or generate data
    │   └── GetParamters.py 	<- downloads parameters from movie name
    │   └── FactorQuery.py 	<- downloads paramter info and finds movie ratings
    │   └── LineCount.py 	<- counts the lines in a file, used for searching IMDB ratings
    │   └── Search.py 		<- searches the IMDB rating files with either IMDB id or TMDB id
    │   └── SaveLoadJson.py 	<- standardizes the saving and loading of Json data between scripts
    |
    ├── models         <- Scripts to train models and then use trained models to make
    │   │                 predictions
    │   └── Stats.py <- generates some guess based on ratings.txt
    │
    └── visualization  <- Scripts to create exploratory and results oriented visualizations
        └── visualize.py