Classic Time Series Forecasting Method

Table of Contents:

Machine learning methods can be used for nomenclature and predictions for time series problems. Before delving into machine learning methods for time series, it's a good idea to check if there is an old classic linear time series prediction method. Classical time series prediction methods can focus on linear relationships, but they are sophisticated and perform well on a wide range of problems. The data are well prepared, and the methods are well structured.

The 11 different classic time series prediction methods are as follows:

1. Autoregression (AR)
2. Moving average (MA)
3. Autoregressive moving average (ARMA)
4. Autoregressive integrated moving average (ARIMA)
5. Seasonal autoregressive integrated moving average (SARIMA)
6. Seasonal autoregressive integrated moving average (SARIMAX) using exogenous regression variables
7. Vector autoregression (VAR)
8. Vector autoregressive moving average (VARMA)
9. Vector autoregressive moving average (VARMAX) using exogenous regression variables
10. Simple exponential smoothing (SES)
11. Holt-Winters exponential smoothing (HWES)

Autoregression (AR)

The autoregressive (AR) method models the next step in a sequence as a linear function of the observations in the previous time step. The notation for the model involves specifying the order of the model p as a parameter to the AR function: AR (p). For example, AR(1) is a first-order autoregressive model. This method is suitable for univariate time series without trend and seasonal components.

Moving Average (MA)

The moving-average (MA) method models the next step in a sequence as a linear function of the residual error of the midpoint process at the previous time step. The moving stereotype model varies from moving stereotypes in a time series. The notation for the model involves specifying the order of the model q as a parameter to the MA function: MA (q). For example, MA (1) is a first-order moving stereotype model. This method is suitable for univariate time series without trend and seasonal components.

You can use ARMA admissions to create an MA model and establish a zero-order AR model. You need to specify the order of the MA models in the order argument.