Time sequence prediction method based on time convolution and LSTM

Abstract

The invention discloses a time sequence prediction method based on time convolutions and LSTM (Long Short Term Memory Network). The method includes: obtaining sample data of a time sequence predictionproblem, wherein the data include time sequences and prediction target sequences; using different hyperparameter combinations to establish multiple candidate models based on a TC-LSTM model; preprocessing the time sequences to obtain window sequences and a target set, and obtaining a training set and a test set through dividing; using the training set to train the candidate models; using the trained candidate models to predict a window sequence subset in the test set, and calculating a root mean square error of a prediction result of each candidate model and a subset of the target set in thetest set; selecting a candidate model of a smallest root mean square error to use the same as a prediction model; and preprocessing a to-be-predicted time sequence to obtain to-be-predicted window sequences, and then using the prediction model to predict the to-be-predicted window sequences to obtain prediction target values. The method can improve accuracy of time sequence prediction.

Description

technical field [0001] The invention belongs to the field of deep learning and time series prediction, and more specifically, relates to a time series prediction method based on time convolution and LSTM. Background technique [0002] Time series is a different data representation of a phenomenon over time, so it is a series of linear sequence data. Time series forecasting is to analyze, process, discover its changing rules and predict future trends of this type of data. method. Time series forecasting is widely used and is closely related to various scenarios in real life, such as short-term and long-term forecasting of commodity sales, and analysis of stock price fluctuations in financial markets. [0003] The traditional time series prediction is mainly based on the ARIMA (Autoregressive Integrated Moving Average Model, ARIMA) model. The ARIMA model converts the non-stationary time series into a stationary time series, and then the dependent variable is only affected by ...

AI Technical Summary

Problems solved by technology

[0006] In view of the defects and improvement needs of the prior art, the present invention provides a time series prediction method based on tim

Images