Training: The model is trained using a dataset of contracts with known outcomes (malicious or benign). This phase involves adjusting the model's parameters to minimize error and improve predictive accuracy. (Forta Dataset)
Current results:
Precision: 63% Recall: 79% F1-Score: 70%
Validation: A separate dataset is used for validation. Techniques like cross-validation help in tuning the model and selecting the best version to prevent overfitting or underfitting.
Testing: The final step involves evaluating the model's performance on a test dataset that it has never seen before. This provides an unbiased assessment of its predictive capabilities.