Question 18: For a binary classification problem (e.g., pred…
Question 18: For a binary classification problem (e.g., predicting bad loan vs. good loan), Model A has an AUC of 0.82 and Model B has an AUC of 0.65. A student argues that we do not need to specify a cut-off probability to decide which model has better overall discriminatory power. Is the student correct? Why or why not?
Read DetailsPart II. Short-Answer Questions. Problem I. Questions 16-20…
Part II. Short-Answer Questions. Problem I. Questions 16-20 are based on the below. Consider the following statement: “When fitting a neural network regression model, if we increase the number of hidden nodes from 3 to 10, the model will always perform better on new, unseen test data because it has more capacity to learn complex patterns.” Question 16: Do you agree or disagree with the above? Please provide a short sentence for your reasoning.
Read Details