Hello Since the test data integrated in the context file is all NaN, one might ask for the utility of the features. So the question is Are we supposed to treat the problem as a simple time series prediction if so what is exactly the utility of the featured data. Else are we supposed to predict features and after that predict the target ??