Temperatures data preprocessing for short-term gas consumption forecast

Final quality of every prediction model is significantly dependent on the quality and good preprocessing of its input data. A case-based reasoning method is presented for outside temperature data preprocessing for the purpose of short-term gas consumption forecasting. The great advantage of our approach is its simplicity, stability and predictability of results. The suggested method is used both for handling missing data and for user input preprocessing when hourly temperatures profile is computed from few input values. The presented algorithm is computationally robust and is implemented in several real-time systems.