Hi Surbhi! Thanks for the question! Consider that the test data was already part of the original dataset, that’s how I ran testing, the temperature was available at any time. It’s not “real forecasting” yet. For that I’d have to ask the authors of the dataset more data from the first months of 2018 and see how the model behaves. Hope this answers your question! If not, please let me know if you have any other doubts!