问题
H2O's documentation doesn't provide clear definitions for each column in the gains/lift table output. I'm not sure how the capture rate is being calculated, and there is a score column that is not mentioned in the documentation.
Here's what the output looks like.
The raw java file is here -- I tried finding the answer to my question in there but had difficulty making sense of it. Thanks.
回答1:
The capture rate is the proportion of all the events that fall into the group/bin. E.g. if 90 out of total 100 positive outcomes/events fall into the first bin, then the capture rate for that bin is 0.9.
回答2:
For a given row x in the table:
cumulative_data_fraction = number in group x / all observations
lower_threshold = minimum of predicted probability in group x
lift = response_rate / overall response rate (i.e., cumulative_response_rate in group 16)
cumulative_lift = cumulative_response_rate / overall response rate
response_rate = percent of responses in group x
score = average of predicted probability in group x
cumulative_response_rate = response rate for all observations in groups 1 through x
cumulative_score = average of predicted probability for all observations in groups 1 through x
capture_rate = percent of all responses that are in group x
cumulative_capture_rate = percent of all responses that are in groups 1 through x
gain = (lift - 1) * 100
cumulative_gain = (cumulative_lift - 1) * 100
来源：https://stackoverflow.com/questions/52521844/what-are-the-column-definitions-for-h2os-gains-lift-table