In refactored code, predict the times and marks, evaluate the predictive performance

Hi, in the refactored version of the code, how can I predict the occurrence times of the next n events and forecast the types of these events (marks)? How can I evaluate the predictive performance on the same dataset across different models? Would it be possible to provide code for the questions above? Thank you very much.