Research on Multistep Time Series Prediction Based on LSTM
Yupeng Wang, Shibing Zhu, Changqing Li · 2019 3rd International Conference on Electronic Information Technology and Computer Engineering (EITCE) · 2019
LSTM (Long Short-Term Memory) is a neural network model that can effectively predict time series. This paper studies the problem of LSTM multi-step time series prediction. By studying and comparing two methods of multi-step input and seq2vec, the paper provides reference for LSTM in the field of time prediction. Firstly, the basic theory of LSTM is introduced. Then the LSTM model is used to analyze and predict the selected data set. The result proves that LSTM can effectively predict seasonal time series data in multiple steps. At the same time, the paper summarizes and presents the advantages and disadvantages of the two methods.