COS781 Project - Steam Data Analysis and Recommendation System

This GitHub repository hosts the code and resources for a project that analyzes Steam user behavior data and builds a recommendation system using various techniques. The project is designed to explore and visualize the dataset, generate recommendations, and evaluate the effectiveness of different recommendation methods.

Features

1. Data Cleaning and Exploration

• Loads and preprocesses a dataset of Steam user behaviors.
• Cleans missing values and removes irrelevant columns.
• Provides summary statistics and insights.

2. Data Visualization

Bar plots of:

• The top 10 most played games.
• Games with the highest average playtime.

3. Recommendation Systems

Implements three recommendation approaches:

• Collaborative Filtering: Recommends games based on cosine similarity of user behaviors.
• Association Rule Mining: Identifies frequent itemsets using Apriori and computes association rules.
• Matrix Factorization: Applies Singular Value Decomposition (SVD) to recommend games.

4. Evaluation

Evaluates recommendation methods using:

• Precision
• Recall
• Mean Average Precision (MAP)

Installation

Prerequisites

• Python 3.12.4 or higher.
• Jupyter Notebook for running the .ipynb file.

Dataset

The project uses the steam-200k.csv dataset, which contains Steam user behavior data, including user interactions such as purchases and playtime for various games. The dataset is loaded using pandas.

Dataset File: steam-200k.csv

Delimiter: ;

Usage

Running the Code

1. Ensure the dataset is in the repository directory or adjust the file path in the code.

2. Open the notebook Project.ipynb and execute cells sequentially to:

• Clean and explore the data.
• Visualize game statistics.
• Build and evaluate recommendation systems.

Outputs

Visualizations:

• Top 10 most played games.
• Top 10 games with the highest average playtime.

Recommendations for users and games using three techniques.

Evaluation metrics (Precision, Recall, MAP) for all methods.