TD_ARIMAESTIMATE estimates the coefficients corresponding to an ARIMA (AutoRegressive Integrated Moving Average) model, and to fit a series with an existing ARIMA model. The function can also provide the "goodness of fit" and the residuals of the fitting operation. The function generates model layer used as input for the TD_ARIMAVALIDATE and TD_ARIMAFORECAST functions. This function is for univariate series.
TD_ARIMAESTIMATE takes one or two inputs, the second input is optional. The first input is a time series. The second input references the model context. When only one input is passed in, TD_ARIMAESTIMATE operates in a coefficient estimate mode. When two inputs are passed in, TD_ARIMAESTIMATE operates in a model apply mode. When the second input is passed in, you must include an INPUT_FMT(MODE()) clause.
You can use the FIT_PERCENTAGE parameter to pass a portion of the data, such as 80%, to the TD_ARIMAESTIMATE function. The ART produced includes the ARTVALDATA layer which contains the remaining 20%, and can be used with TD_ARIMAVALIDATE for the validation exercise.
- TD_BREUSCH_GODFREY, serial correlation test
- TD_DURBIN_WATSON, serial correlation test
- TD_BREUSCH_PAGAN_GODFREY, heteroscedastic variance test
- TD_WHITES_GENERAL, heteroscedastic variance test
- TD_GOLDFED_QUANDT, heteroscedastic variance test
- Run the TD_ARIMAESTIMATE function to estimate the coefficients for the ARIMA model.
- [Optional] Run TD_ARIMAVALIDATE function to validate the ARIMA model, when FIT_PERCENTAGE was not set to 100 percent in TD_ARIMAESTIMATE.
- Run the TD_ARIMAFORECAST function with input from step 1 (FIT_PERCENTAGE was set to 100 percent) or step 2 to (FIT_PERCENTAGE was set to a value less than 100 percent).