- This repo contains backend machine learning based scripts used for game analysis and recommendation for the project GamerHood
- GamerHood, a web application built with PySpark(Python) (WIP) for game analysis and recommendations. This enables users to access game information and related content and, based on their behaviour, get suggested games. This utilizes the help of MongoDB Atlas for managing databases and algorithms like clustering, regression, recommendation engine for analysis.
- This repo also contains scripts used for scraping and cleaning all game related information from Steam Official Website
- Compiled Dataset can be found here: 80000 Steam Games DataSet
- Machine Learning Part has been implemented using PySpark and Database Handling part has been implemented using MongoDB Atlas
- Recommendation using Alternating Least Squares (ALS) has been implemented on Game Data and Game User Data using PySpark
- Dynamic Determination of Appropriate number of Clusters for K-Means Clustering using Elbow Method has been implemented
- Linear and Decision Tree Regression has been implemented for review data of Among Us Game. View Plot
- KMeans Clustering has been Implemented for games related to GTA V with respect to similar popular tags. View Plot
- Clustering Parameters Used
- Number of Positive Reviews
- Number of Negative Reviews
- Price
- Clustering Parameters Used
- WordCloud has been implemented for Fall Guys: Ultimate Knockout Game