Classification Project - Predicting Win/Loss rate of League of Legends & Understanding the Patch Notes through Data Findings
The goal of this project is to predict the win/loss rate of the ranked match games in a game called League of Legends and potentially suggest which features are determining factors that influences the game. Furthermore, with the key findings I try to understand an important patch notes that were done that may have influenced the meta of the game for all players. The dataset consists of 26,000+ rows of Challenger data from Kaggle. This project utilizes classification algorithms such as KNN, Logistic Regression, and Random Forest. Although Random Forest provided the most accurate result, Logistic Regression was chosen as the final model for better interprebility.
- ROC/AUC Curve for 3 Different Models
- Confusion Matrix
- Presentation PPT
- Presentaiton Recording - YouTube
- Python(Pandas + Numpy) for data manipulation
- Sklearn package for Logistic Regression model (final chosen model)
- F1, AUC/ROC scores for classification metric
- Matplotlib for visualizations
- Confusion Matrix to convey result