I am creating a visualization that displays the aggregated results of LIME SubmodularPick, basically I plot a histogram of the result as well as a boxplot. However when I look at the table I notice a lot of values are missing, instead they appear as NaN
Why are there missing values?
Is a missing value due to the fact that the feature wasn't considered in the explanation or was it not important?
If there are all NaN and a few really high scores (0.99) does it mean that the feature is important only in a few cases?
