Closed m-pauper closed 1 day ago
Hi. As you found, my parameter description is not accurate. The fraction
should be the valid quantitative fraction (1 - missing ratio). In fact, a larger fraction
corresponds to a stricter filtering. Thank you for your observation, and I will modify the description of parameter fraction
in future updates.
Thanks for the quick reaction!
Hello, thanks for publishing DEP and DEP2.
I just wanted to give some feedback on
filter_se
's parameterfraction
as I find it a bit misleading. The description says:"A numeric from 0 to 1, threshold of missing occupancy of each row"
However, setting the threshold at e.g. 0.4, means that a feature can have 60% missing values. I would expect it to mean that 40% missing values is the maximum threshold allowed.