| 摘要 |
PROBLEM TO BE SOLVED: To provide a method for adaptively predicting time series data, which can perform highly accurate prediction even when prediction object time series data show unexpected tendency without performing unneeded calculation and also brings about both the improvement of predictive accuracy and the realization of prediction reason explanation. SOLUTION: This method for adaptively predicting time series data, which calculating a predicted value about future by using the past time series data, constructs a prediction model by inputting/accumulating data for prediction model construction, obtains a predicted value by conducting prediction according to the constructed prediction model, corrects the predicted value by calculating a correction coefficient on the basis of a prediction error calculated by using the past time series data and notifies the calculation grounds of the predicted value. |
