Considering we have a data set of 944 elements. One with 45% valid instances (433 instances) and the other with 73.8% valid instances (697 instances). The F1 score for the first group with lower valid instances is higher than the other. Is there any statically measurement that includes the number of instances into the consideration?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.