Given that this is a forecasting competition I have a question regarding historical weather data. The data provided in open data sources are observed weather data. Ideally, we would like to have past weather forecasts but I did not manage to find any dataset.
So the question is: is it OK to use observed weather data for the end of December 2018 while forecasting?
Thank you for your question. You are totally right; in the real-world we would need to forecast the weather data, but since we are dealing with historical data, you can assume you have a perfect weather forecast, and hence, use observed weather data. Please note that the data is for 2018, and thus you’d better use the weather data for 2018.