Question

A machine learning engineer trains a random forest classifier to predict equipment failures. On the training set the model achieves 98% accuracy, but on the validation set it achieves only 74% accuracy. A learning curve shows the training score staying very high while the validation score plateaus well below it, with a large persistent gap between the two curves. The engineer wants the most effective way to improve validation performance. Which action is MOST likely to reduce this gap?

Accepted Answer

Reduce model complexity by constraining tree depth and add more training examples if available.

Answer

Increase the maximum tree depth and the number of estimators to give the model more capacity.

Answer

Switch the evaluation metric from accuracy to F1 score to better capture performance.

Answer

Lower the classification threshold from 0.5 to 0.3 to capture more positive predictions.