Some prediction features contain categories that were not present in the training features. These missing categories were not used in the prediction at these locations. The number of locations that ignored at least one categorical variable is displayed in the message.
No action is required. However, consider including training features that include all categories of the prediction features.