search

FateMurphy / CEEMDAN_LSTM

CEEMDAN_LSTM is a Python project for decomposition-integration forecasting models based on EMD methods and LSTM.
MIT License
217 stars 42 forks source link

Prediction Method #36

Open gaochenxiai opened 3 months ago

gaochenxiai commented 3 months ago

Dear author, how does your prediction section work? Do you predict the next value using the previous actual value, or do you use the already predicted values?

FateMurphy commented 3 months ago

My other repository CEEMDAN-VMD-GRU might be easier to understand. Most functions in this module use the previous actual value for one-step-ahead forecasting, unrelated to the test set length. However, some methods have a step of fitting which will use the interprocess predicted value. Or, some functions like rolling_predict() can reinput the predicted value to forecast several timesteps.