Using Data to Build the (Statistically) Perfect Oscar Film
The skills the authors demonstrated here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Introduction
The Academy Awards, colloquially known as the Oscars, are prestigious film movie awards. These are given for a variety of reasons and these awards are so highly valued by those who produce movies, that many movies are produced specifically to target these awards based on previous data. These movies are colloquially called "Oscar-Bait" movies. They are typically lavishly produced, period-dramas, often set to the backdrop of a major world calamity or at least a historically documented calamity (WWI, WWII, a Cholera outbreak, The Spanish Plague, etc.), often involving cinema or some form of entertainment industry. This occurs based on a perception that these types of movies are historically most likely to win Oscars.
While a great many of movies that fall under this category are terrific, not all of them are Oscar-Bait movies as this has a connotation of being artificial, set-up to win an award rather than naturally being a thoughtful movie. As a result many movies that follow this formula end up not nominated for even one Academy Award.
Therefore, we would like to find a way to create Oscar-winning film plots, by assembling them based on the statistical correlation of different tropes appearing in Oscar Winning movies.We do this by scraping the website TvTropes, a pop culture trope analysis website, that compiles a list of tropes that they have collected and found in each movie. By cataloguing the tropes found in Oscar Winning Movies, we can see if there is any merit to using a trope-based analysis to construct statistically successful movies.
Tropes are not as simple as to compare blindly however. The assumption made henceforth is that the tropes are more meaningfully distinguished when comparing various movies with shared genres. Without any shared genre, its unlikely that the shared tropes are those presented or used in a similar way. In order to reduce null results, we stipulate that the movies share at least one genre.
Yearwise Analysis Using Data
Genrewise Analysis Using Data
- Introduction, in which you should describe the following: what the Oscars are in brief, the problem of "Oscar-Bait" movies--How they can be immediately spotted. Therefore, there is need for a new approach--we do a trope based analysis of the Oscar movies using a free online public resource data: TVTropes. First, we find the highest grossing Oscar Winners, see what differentiates the stories of the Winners VS the Nominees using the data --We focus our analysis to targeting two genre's out of type--Drama (all Oscar winners are this), Crime (which rarely wins Oscars), and Action. Then, we are trying to design the plot of an Action Crime Drama movie set up to win the following Oscars-- 1. Best Picture, 2. Best Director, 3. Best Actor, 4. Best Actress, 5. Best Screenplay, also known as the "Big 5" Academy Awards.
2. Method, in which you should describe: in brief, some of the tools, strategies, and data used, some challenges (like querying the website) and how you overcome them. Understanding how the site is organized, figuring out how to pull the right information.
3. Trope Analysis--Now this is going to be broken down into pictures. Have a story building for each of these--Highest Award Winning, Highest Grossing, for each of the Action and/or Crime movies for the Big 5. See which tropes are most popular for each of these and provide your own analysis by checking out he movie and adding your own twist. You should have a lot of subsections and a lot of clear graphs and data, consider switching some of this code to R or really swagging out the Jupyter Notebook. Idk, but dash is not very good. E.x.
1.Best Picture--a. Nominated but not Won vs Won Trope analysis. b. Highest Award Winning in Best Picture Winners compared to Highest Grossing Best Picture Winners. We do this analysis for all 5 awards.
2.Inverse Trope Analysis 2--Now we synthesize the knowledge taken from all these graphs. We break down the Highest Grossing Tropes, The Highest Award Winning and we come up with two or three example stories with some of the tropes--We now analyze the Bottom Gross--and see what tropes they had that were so unpopular. Now from all of this we can have like 5 Story examples of ideas that might have a really good responsiveness with Oscar Audiences and also Highly Grossing. Need to Have Word Clouds comparing the Good vs the Bad (Green vs Red).
Conclusion--Some final thoughts on the project limitations, places to improve the current project in its current application to movies (more sophisticated analysis of tropes, more detailed sentiment analysis of each of them, more details about the movie--maybe summarizing plots so that we can extract story beats and compare those), separate application of the trope analysis method--could apply trope-base analysis to other things on TVTropes--like Anime/Manga, TV Shows, Other types of Content. Business Applications could include using data collected by users of a company that wants to produce content--content creators who want to learn what tropes are trending in their genre, which are critically acclaimed and which are losing audience favor-ability.
About Author
