Partial dependence plots are empty, partial dependence shows null list

Question

I'm trying to create partial dependence plots for a classification model (XGBClassifier). I was able to get variable importance, but some of the partial dependence plots for those variables are empty. These are the two functions I have tried:

PartialDependenceDisplay.from_estimator(model, X_data, [variable])
plot_partial_dependence(model, X_data, [variable])

where model is my trained model, X_data is the dataset of variables, and variable is the specific trait that the partial dependence plot is not showing up for.

I also see that when I use the `partial_dependence| function to see the values, there are nan values where there should be numbers:

partial_dependence(model_type[0], X, [variable])

which gives me:

(array([[0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357,
         0.11820357, 0.11820357, 0.11820357, 0.11820357, 0.11820357]],
       dtype=float32),
 [array([nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan])])

when it should give me two arrays of non empty, unique numbers. I've tried a bunch of different things, but I can't figure out how to fix this, or where in my code there is an error that causes this problem.

Partial dependence plots are empty, partial dependence shows null list

0 Answers