Performance of Deep Learning models with transfer learning for multiple-step-ahead forecasts in monthly time series
This work addresses the need for improved forecasting accuracy in monthly time series, which is incremental as it builds on existing deep learning and transfer learning techniques.
The paper tackled the problem of forecasting monthly time series by comparing deep learning models with and without transfer learning against traditional methods, finding that models like TCN, LSTM, and CNN with transfer learning often outperformed traditional approaches, while TCN and LSTM without transfer learning performed similarly or better in some cases.
Deep Learning and transfer learning models are being used to generate time series forecasts; however, there is scarce evidence about their performance prediction that it is more evident for monthly time series. The purpose of this paper is to compare Deep Learning models with transfer learning and without transfer learning and other traditional methods used for monthly forecasts to answer three questions about the suitability of Deep Learning and Transfer Learning to generate predictions of time series. Time series of M4 and M3 competitions were used for the experiments. The results suggest that deep learning models based on TCN, LSTM, and CNN with transfer learning tend to surpass the performance prediction of other traditional methods. On the other hand, TCN and LSTM, trained directly on the target time series, got similar or better performance than traditional methods for some forecast horizons.