Method for bi-directionally processing missing data by using ARIMA-LSTM model

The method for bidirectionally processing the missing data by using the ARIMA-LSTM model comprises the following steps of: preprocessing model training data; a forward ARIMA model and a forward LSTM model are established, and the forward ARIMA model and the forward LSTM model are combined to generate a forward ARIMA-LSTM model; establishing a reverse ARIMA model and a reverse LSTM model, and combining the reverse ARIMA model and the reverse LSTM model to generate a reverse ARIMA-LSTM model; preprocessing the to-be-filled data; the forward ARIMA-LSTM model and the reverse ARIMA-LSTM model are used for completing prediction; fusing and complementing missing data by using the obtained predicted value, and obtaining the predicted value according to a formula # imgabs0 #; according to the method, stable data missing value filling is processed; the LSTM is a non-linear prediction method based on sample training, errors generated when the ARIMA performs missing filling on non-stationary data are corrected, data are