- Yvon Bilodeau
- May 2022
A fictional person is the head of a new movie studio. They are new to the industry, and have asked for recommendations to help them make successful movies.
For this project, I have been requested to extract information from the IMDB (Internet Movie Database) and TMDB (The Movie Database) in order to analyze what makes a movie successful.
For Phase 1 of the project, the stakeholder has requested an extraction of files from the IMDB (Internet Movie Database), and the creation of a repository to contain them.
The IMDB Provides several files with varied information for Movies, TV Shows, Made for TV Movies, etc.
Extraction - From their previous research, they realized they would like to focus on the following files:
- title.basics.tsv.gz
- title.ratings.tsv.gz
- title.akas.tsv.gz
Transformation/Filtering - The stakeholder would only like information included for movies based on the following specifications:
- Exclude any movie with missing values for genre or runtime
- Include only full-length movies (titleType = "movie").
- Include only fictional movies (not from documentary genre)
- Include only movies that were released 2000 - 2021 (include 2000 and 2021)
- Include only movies that were released in the United States
- Movie-Production-Business-Analysis repository
- title_basics.csv.gz
- title_akas.csv.gz
- title_ratings.csv.gz
For Phase 2 of the project, the stakeholder realized that there is no financial information included in the IMDB (Internet Movie Database) data, e.g. budget or revenue.This will necessary in attempting to analyze which movies are successful.
The Movie Database (TMDB) however, is a great source of financial data.
Fortunately, The Movie Database (TMDB) offers a free API for programmatic access to their data.
1. API Data Extraction
- The stakeholder has requested an extraction of the budget, revenue, and MPAA Ratings (G/PG/PG-13/R), also called "Certification".
- The stakeholder has requested the results be extracted and saved for movies that meet all of the criteria established in Phase 1 of the project.
- As a proof-of-concept, they requested a test extraction of movies that started in 2000 or 2001, both years to be saved as both separate and a combined.csv.gz files.
2. Exploratory Data Analysis
- Once the data has been extracted from the API, they would like some initial exploratory questions be answered:
- How many movies had at least some valid financial information?
- How many movies are there in each of the certification categories (G/PG/PG-13/R)?
- What is the average revenue per certification category?
- What is the average budget per certification category?
- final_tmdb_data_2000.csv.gz
- final_tmdb_data_2001.csv.gz
- tmdb_results_combined.csv.gz
The stakeholder has requested that the data that has been extracted and transformed in Phases 1 & 2 of the project, be utilized to create a MySQL database.
The tables should be normalized as well as possible prior to adding them to the new database.
- Note: an important exception to their request is that they would like you to keep all of the data from the TMDB API in 1 table together (even though it will not be perfectly normalized).
- Only imdb_id, revenue, budget, and certification columns need to be kept.
SQL "Movies" database with the following tables:
- title_basics
- title_ratings
- title_genres
- genres
- tmdb_data
- In Phase 2, as a proof-of-concept, the stakeholder requested a test extraction of movies that started in 2000 or 2001. Each year to be saved as a separate .csv.gz file.
- In Phase 4, the stakeholder has requested additional years to be extracted, and years 2000-2021 are to be combined into a single TMDB .csv.gz file.
- final_tmdb_data_2000.csv.gz
- ...
- final_tmdb_data_2021.csv.gz
- tmdb_results_combined_final_df.csv.gz
For Phase 4 of the project, the stakeholder has also requested statistical tests to obtain mathematically-supported answers to their questions:
- Does the MPAA rating of a movie affect how much revenue the movie generates?
- Does the genre of a movie affect how much revenue a movie generates?
- Do movies that are over 2.5 hours have a significantly different revenue than movies that under 1.5 hours in length?
For each question:
- They would like to know if a statistically significant difference exists for each hypothesis.
- They would like to know the p-value of the test.
- They would like a visualization that supports the findings of the test.
- The p-value for the test was 5.504858804917085e-168
- It was < the alpha value of 0.05, so
- A statistical significance exists. The null hypothesis is rejected and the alternative is supported that..
- The MPAA rating of a movie does affect how much revenue the movie generates.
- The p-value for the test was 1.760985049928353e-249
- It was < the alpha value of 0.05, so
- A statistical significance exists. The null hypothesis is rejected and the alternative is supported that..
- The genre of a movie does affect how much revenue a movie generates.
Do movies that are over 2.5 hours have a significantly different revenue than movies that under 1.5 hours in length?
- The p-value for the test was 2.428023688468123e-11
- It was < the alpha value of 0.05, so
- A statistical significance exists. The null hypothesis is rejected and the alternative is supported that..
- Movies that are over 2.5 hours have a significantly different revenue than movies that under 1.5 hours in length.
Summary of the hypotheses tests
- The MPAA rating of a movie does affect how much revenue the movie generates.
- Movies that have an MPAA Rating of PG make the most revenue.
- The genre of a movie does affect how much revenue a movie generates.
- Adventure, Sci-fi, and Action movie genres, in descending order generate the greatest revenue.
- Movies that are over 2.5 hours have a significantly different revenue than movies that under 1.5 hours in length.
Recommendations
In order to maximize revenue for a movie, it would be recommended to produce:
- a PG rated, adventure movie with a runtime greater than 2.5 hours.