Skip to content

This final project uses two machine learning models, XGBoost and CatBoost, to train and predict outcomes on the given dataset.

Notifications You must be signed in to change notification settings

ThanhDang-Vn/ensemble-catboost-xgboost

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Ensemble Learning - XGBoost vs CatBoost

πŸ“‹ Project Overview

Comparative analysis of two powerful Gradient Boosting Decision Tree algorithms:

  • XGBoost: Optimized gradient boosting with regularization and sparsity awareness
  • CatBoost: Ordered boosting with native categorical feature support

This project implements both algorithms on multiple datasets with hyperparameter tuning using Optuna.

🧠 Key Differences

Aspect XGBoost CatBoost
Split Finding Exact greedy / Histogram-based Ordered splits
Categorical Features Requires One-Hot Encoding Native support via Target Encoding
Tree Type Standard Symmetric (Balanced)
Training Speed Fast (numerical data) Moderate
Inference Speed Moderate Fast
Robustness Sensitive to hyperparameters More robust defaults

πŸ“Š Datasets

  • Breast Cancer Classification: Binary classification dataset
  • Real Estate: Regression task for price prediction
  • Additional datasets for comprehensive evaluation

πŸ”¬ Evaluation Metrics

Classification: AUC-ROC, LogLoss, F1-Score Regression: RMSE, MAE Efficiency: Training time, Inference time

πŸ“ Project Structure

  • classification/: Classification notebooks and models
  • assets/: Dataset files
  • catboost_info/: Training artifacts and logs

πŸ“š Key Findings

  • Categorical Data: CatBoost excels with high-cardinality features
  • Hyperparameter Tuning: XGBoost requires careful tuning; CatBoost more stable
  • Trade-offs: XGBoost faster training vs CatBoost faster inference

About

This final project uses two machine learning models, XGBoost and CatBoost, to train and predict outcomes on the given dataset.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •