Question

An online social media platform has asked you to develop a chu propensily rmode:l. Youhav bild a boosied ire model with 500 trees and maximum depth of 50. The AUC value is 0.95 on training data, but only 0.5 on test data. To improve model performance on test data, which step would you recommend to try next? A. Reduce maximum tree depth. B. Increase number of trees. C. Reduce from 10-fold cross validation to 3-fold. D. Apply a logarithmic transformation to features that are not normally distributed.

Answer 1

To improve the model performance on test data, I would recommend increasing the number of trees in the boosted tree model.

Step-by-step explanation:

To improve the model performance on test data, I would recommend trying option B, which is to increase the number of trees in the boosted tree model. Increasing the number of trees can help to improve the generalization ability of the model and reduce overfitting, which may be the reason for the lower AUC value on the test data. By increasing the number of trees, the model can capture more complex patterns in the data and potentially improve its performance on unseen data.