AdaBoost Regression to Select Predictors

Is it possible to use the LSBooost alogorithm to select a subset of appropriate predictor variables in a similar way to the step-wise regression function. I cannot find any examples of using it to select variables (and ideally showing final coefficients?)? Thanks

Answer 1

[Neil Guertin]

The fitrensemble function uses the LSBoost algorithm. Then you can use the removeLearners function to remove members of the ensemble, as described in this example:

https://www.mathworks.com/help/stats/compactregressionensemble.removelearners.html?s_tid=srchtitle#bty1jnv-1

Also, the Regression Learner and Classification Learner apps allow you to use Principle Component Analysis to train on only the most significant predictor variables.

Answer 2

[Daniel Terry]

thanks Neil - much appreciated