CSYE7200 - Scala_FinalProject

Game Recommendation System

Group Members:

• Chen Ye - 002951477 @yechen0704
• Xinzhuo Liu - 002197134 @ZhongLBuL
• Jingru Xiang - 001586653 @RubyXiang

Outlier:

Data Source:

• User game association table: https://www.kaggle.com/code/danieloehm/steam-game-recommendations/data
• Game rating association table: https://www.kaggle.com/datasets/nikdavis/steam-store-games

Methodology :

• Scala
• Spark
• Collaborative Filtering
• Swagger UI
• Play Framework
• MongoDB

Dependencies:

• Scala v2.12
• Spark v3.1.3
• sbt v1.6.2
• MongoDB v4.4, Mongodb Compass(uri: "mongodb://localhost:27017/testdb")

(Our project only needs to use sbt to manage external packages and run all parts)

Instruction:

Firstly, you need import Scala_FinalProject, collaborativeFiltering, dataPreprocess, play, recommendation directories as modules when you start the project to run unit test, and then:

1. Data Preprocess: 'cd dataPreprocess', in 'dataPreprocess' directory, use 'sbt run' to start. When finished, 'game.collection', 'train.collection', 'validation.collection', and 'test.collection' can be seen in Mongodb Compass.
2. Train Model: 'cd collaborativeFiltering', in 'collaborativeFiltering' directory, use 'sbt -v -J-Xmx2048m run' to train and validate data from Mongodb. When finished, the directory 'myModel.model' will appear in 'collaborativeFiltering' directory, and 'validPred', which stores the prediction values, is created in MongoDB.
3. Static Prediction: 'cd recommendation', in 'recommendation' directory, use 'sbt run' to create prediction collection 'testPred' for test collection.
4. Start System & Dynamic Prediction: 'cd play', in 'play' directory, use 'sbt -v -J-Xmx2048m run' to start project. Then, to input 'http://localhost:9000/' on web page to use our system.