dsawali / movies-project-analysis

Data science project that analyzes movie datasets from WikiData, and Rotten Tomatoes

README

Dependencies

Make sure to have Pandas, NumPy, and PySpark installed before running any of these

To get started, you will need these datasets from the movies folder in the cluster:

• wikidata-readable that is generated from build_useful_movies.py
• label_map
• rotten-tomatoes.json.gz

Important Files

• build_useful_movies.py
• correlations.py
• analyze.py
• profitable_time.py

build_useful_movies.py

run command: spark-submit build_useful_movies.py wikidata-movies label_map output

Cleans the data output from build_wikidata_movies.py and maps all wikidata ids to its respective labels in label_map

outputs: movies-readable.json.gz

correlations.py

run command: python3 correlations.py

Loads movies-readable.json.gz and rotten-tomatoes.json.gz into the program, and outputs the correlations of specific columns in the dataset into the terminal using a bunch of print statements. Summary of the results can be found in the Project Summary

analyze.py

run command: python3 analyze.py

Loads movies-readable.json.gz and rotten-tomatoes.json.gz into the program again, and takes a deeper look into the data. Mainly focused on cast_member, director, and made_profit. Outputs four graphs that gives insight on movie profit and ratings.

outputs:

• top_directors.png
• top_actors.png
• directors_sd.png
• actors_sd.png

profitable_time.py

run command: python3 profitable_time.py

Loads movies-readable.json.gz and rotten-tomatoes.json.gz, and groups the movies/works that made profit by publication month. Outputs a histogram on what publication month had the most movies that made profit.

output: month-profit-count.png

Note that both movies-readable.json.gz and rotten-tomatoes.json.gz should be in the same folder as the Python programs above