This project involves training and evaluating two machine learning models with different feature manipulation strategies. The goal is to compare their performance and assess their effectiveness in a specific task.
- Model01: This model was trained using features with minimal manipulation or correction.
- Model02: This model utilized heavily manipulated features, with ideas primarily sourced from another Kaggle platform user.
- Model01: 0.954545
- Model02: 0.856459
- Model01: 0.76555
- Model02: 0.77272
Despite Model01 achieving a higher accuracy on the training data, Model02 performed better on the leaderboard. This suggests that the feature manipulation in Model02 may have improved its generalization to unseen data, despite a lower accuracy in training.
To replicate or extend this project, ensure you have the necessary dependencies installed and follow the provided scripts for training and evaluation. For detailed instructions, refer to the Installation and Usage section.
- Clone this repository.
- Install the required dependencies listed in
requirements.txt. - Run the training scripts to train the models.
- Evaluate the models using the provided evaluation scripts.
- For further details or questions, please refer to the documentation or open an issue in the repository.