enum class TStepwiseMethod : unsigned int;
Stepwise regression optimization algorithm.
Values
| Name | Description |
|---|---|
| swBackward | Additional variables will be excluded from the existing model, if the quality criteria is improved. |
| swExhaustive | All variable combinations will be computed and the set with best result (according to the criteria) will be selected. This method allows maximum 15 variables. Variable count of more than 15 can quickly result in unresonably large memory and CPU requirements which increases by 2x for each additional variable. |
| swForward | Additional variables will be included in to the existing model, if the quality criteria is improved. |
| swStepwise | Perform a single regression step, show the result with statistics and allow the user to modify the list of included variables before the next run. |