I've been working for almost a year on electricity load forecasting in collaboration with some climate scientists, using temperature data obtained from models. Instead of using directly temperature values, we use the 'anomaly' from the annual average (called 'climatology'). It seems, in climate sciences, that is good practice to use 'cross-validation' to calculate this anomaly: the anomalies of year X are calculated considering the climatology of all the other years except X, a kind of 'K-FOLD' method with K equals to the number of years considered in the whole calculation. I'm a computer scientist and we are writing for an engineering journal and I was wondering if this whole procedure to calculate anomalies with this kind of 'K-FOLD' it is really considered necessary to make our entire work more robust. I've never read of similar procedure in other forecasting or engineering papers and for this reason I'd like to avoid unnecessary work (and code writing) at this stage of the work. Thank you for your comments,


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.