ESRB Rating Machine Learning Application

Problem Summary

A fictional company is currently in development for an upcoming video game. The development team is unsure of the content the game should have in it as it will affect the ESRB rating. The ESRB rating will also determine the audience the marketing team should focus their strategy on to maximize sales for the product when it is released. Due to the ambiguity of the decision-making process for this upcoming title, it is paramount a solution is created to minimize risk and maximize profit. In the proposed solution to these issues, the tool will allow an employee to input selected features the game may have, it will take this data and present a predicted rating the game will most likely have. Upon having the predicted rating, the development can decide their direction and implement the features that will allow their rating to be maintained. Additionally, the marketing team will have a clear vision of who their strategy will target for advertising.

Application Benefits

The application benefits the company as it provides a predicted ESRB rating for the current video game project in development. The development team will be able to explore features and decide which content features will determine the rating they desire. The application allows for multiple content options to be selected and offers an accurate prediction based on previously analyzing collected data from thousands of video game samples. The product fits to fill a knowledge gap among the development and marketing teams. These teams will work more efficiently when they have a projected rating for the game they intend on releasing.

Application Description

The application is based on a classification predictive machine learning model. The model was trained with thousands of samples that include content features and ESRB ratings. This model was exported onto a web-based flask environment hosted. The web application will allow an employee of a fictional company to interact and select several features that the upcoming title may have and output a predictive rating. The page also includes several visual figures to describe the data the model was trained with.