Video Game Market ML Analysis

Summary

A 2025 machine-learning paper merging two Kaggle datasets of video game ratings and sales to compare logistic regression, random forest, neural networks, and gradient boosting.

Problem

Reduce review and platform bias by using sales, scores, and encoded features to classify whether games are top performers.

Role

Designed the analysis, merged datasets, weighted important features, compared algorithms, and wrote the final report.

Process

• Merged a 14,802-game ratings dataset with an 11,493-game sales dataset.
• Encoded publisher, platform, genre, regional sales, global sales, and score features.
• Compared model performance using F1 score and feature importance.

Outcome

Reported F1 scores of roughly 0.87 for logistic regression, 0.92 for neural network, and 0.97-0.98 for random forest.

Tools