Kevin Chng
The blooming of machine learning implementation, it has raised interest from different industries to adopt it for classification and forecasting on time series problem. Before exploring machine learning methods for time series, it is good idea to ensure you have tried classifical and statistical time series forecasting methods, those methods are still performing well on a wide range of problems, provided the data is suitably prepared and the method is well configured. In this article, it listed some classical time series techniques available in MATLAB, you may try them on your forecasting problem prior to exploring to machine learning methods. It give you hints on each method to get started with a working code example and where to look to get more information on the method.
Overview: This article demostrates 11 different classical time series forecasting methods, they are
- Autoregression (AR)
- Moving Average
- Autoregressive Moving Average
- Autoregressive Integrated Moving Average (ARIMA)
- Seasonal Autoregressive Integrated Moving-Average (SARIMA)
- Seasonal Autoregressive Integrated Moving Average with Exogenous Regressors (SARIMAX)
- Regression Model with ARIMA Error
- Vector Autoregression (VAR)
- GARCH Model
- Glostan, Jagannathan and Runkle GARCH Model
My other revelevant articles:
- VAR Model To Predict Malaysia/U.S. Foreign Exchange Rate https://www.mathworks.com/matlabcentral/fileexchange/71767-var-model-to-predict-malaysia-u-s-foreign-exchange-rate
- Stock Prediction Using ARIMA https://www.mathworks.com/matlabcentral/fileexchange/68576-stock-prediction-using-arima
- GDP Prediction Using ARIMA and NAR Neural Network https://www.mathworks.com/matlabcentral/fileexchange/68389-gdp-prediction-using-arima-and-nar-neural-network