differnt baseline scores for the same y

[LiuYYSS]

Hi, if my understanding is correct, the calculation of the baseline score does not involve X. Therefore, for the same Y, even if different X are used, the baseline score should be the same. However, I recently came across a strange phenomenon when using pps.matrix(). As shown in the attached image, three different baseline scores appeared when calculating the baseline score for Y=5. I have uploaded a pickle dump of my pandas dataframe. I hope you can test it and tell me if you can reproduce this error.

dataframe.zip

import ppscore as pps
pps.matrix(dataset_df)

ppscore=1.30
pandas=1.5.3

[alessandro-lica-DO]

Hi, perhaps it could depend on any null values you have in your dataset. I think that the baseline model only uses the samples of the dataset where the x is defined.