The reason why there is no provided test.csv is because it is a forecasting challenge so you don't need a labelled test set you just need to forecast the values for the next 2 weeks. The main question I also cannot answer is whether it is the cummulative clicks or just that specific week day clicks

But from the the objective given:

The objective is to accurately predict the “clicks” a client’s ad receives, one and two weeks into the future

so it is not the cummulative clicks but that day's clicks, based on my own understanding

Thank you, your effort means alot. So, is your csv file submission has ID, target and 5 features or just ID and target?

Just the id and target , you get the features from the feature importances of your trained model

If not cumulative, then would that mean we have two targets, one for first week and the other for second week, which it would be inconsistent with sample submission structure.

Yes you have two targets for the first week and second week and the sample sub is given in that format. It has specific dates you should predict for

Thank you... You're a life saver.