I've been working on time series forecasting using models like LSTM, ARIMA, and Prophet. Recently, I came across Orange Tool, which seems very user-friendly and powerful for visual, low-code machine learning workflows. However, I couldn't find much documentation on how to perform future predictions with Orange — especially for time series data. I'd like to use Orange to forecast future values, such as temperature and precipitation for upcoming years based on historical data. (The dataset contains daily values for parameters like T2M, T2M_MAX, T2M_MIN, PRECTOTCORR, PET, Scaled_PRECTOTCORR) The dataset ends on 2025-04-20, and my goal is to predict the next 5 years for each parameter listed above. Dataset that I use with example values: My questions: Is it possible to use Orange Tool for forecasting future values in time series data? If yes, how should I structure my dataset for prediction? Do I need to generate future dates manually and leave target columns empty? Or should I transform the time series into a supervised format (e.g., lag features)? Are there any widgets or add-ons in Orange that support forecasting or time series modeling? I'd really appreciate any guidance, best practices, or examples on how to make future predictions using Orange. Thank you in advance! Continue reading...
