Question

How to define/select metrics?

Answer 1

There isn’t a one-size-fits-all metric. The metric(s) chosen to evaluate a machine learning model depends on various factors:

1. Is it a regression or classification task?
2. What is the business objective? Eg. precision vs recall
3. What is the distribution of the target variable?
4. There are a number of metrics that can be used, including adjusted r-squared, MAE, MSE, accuracy, recall, precision, f1 score, and the list goes on.