Decomposition and CNN based time series forecasting methods

Time series forecasting is usually based on historical observations and is used to predict trends and values of data for a certain period of time in the future. The usefulness of time series forecasting lies in its ability to reveal the intrinsic patterns in the data, such as trend, seasonality and periodicity, which are of great significance to people's decision making, planning and risk management. In this paper, the current research status of time series forecasting is sorted out, and the feature extraction methods involved in time series forecasting as well as the related theories such as deep learning are deeply analyzed, and the advantages and disadvantages of each method are compared, and on the basis of which, the future development trend of time series forecasting methods is outlooked.