17k_mobile_strategy_games

CS534 Final project.

Intruduction

This dataset is from Kaggle.

Our task is to:

• Figure out what factors contribute to the success of strategy games;
• Predict the number of ratings of testing data.

Dataset

The dataset consists of 17007 games wiht 16 features. The target is to predict the output of average user rating from 0.5 to 5.0.

Requirement

• Python 3
• See requirements.txt.

Models

• Logistic Regression (baseline)
• Boosting tree
• SVM
• CNN
• fastText

Metric

• auroc

Examples

$ sh run.sh lr
$ sh run.sh svm
$ sh run.sh bt

use classification_all.py to plot roc curves of three models in one image.

Results

See cs534_Final_17k_revised.pdf.