Preprocessing notebook has the details of preprocessing while in the model building notebook preprocessing steps applied without verbose.
- Scaling amount and time features
- Random Under-Sampling
- Anomaly detection & Removing outliers
- Dimensionality Reduction and Clustering
- Logistic Regression
- Support Vector Machine
- Decision Tree + SMOTE
- Decision Tree with undersampled data
- AdaBoost
- GradientBoosting
- NN-Perceptron
| Model | Accuracy | Precision | Recall | F1-Score |
|---|---|---|---|---|
| Logistic Regression | 0.926 | 1.0 | 0.847 | 1.0 |
| Decision Tree | 0.873 | 0.905 | 0.826 | 0.904 |
| Support Vector Machine | 0.952 | 0.989 | 0.919 | 0.989 |
| AdaBoost | 0.915 | 0.931 | 0.890 | 0.931 |
| GradientBoosting | 0.915 | 0.951 | 0.868 | 0.951 |
| NN-Perceptron | 0.957 | 0.936 | 0.979 | 0.957 |
Model building without preprocessing notebook is there for you to see and don't use it.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent