Zastosowanie wstępnego uwarunkowania zmiennej objaśnianej do selekcji zmiennych
The application of pre-conditioning of explanatory variable for feature selection

Summary/Abstract: Paul et al. [2008] proposed pre-conditioning of explanatory variable that consists of two steps. Firstly, supervised principal components are applied for prediction and then regression model is constructed again. This time the response is replaced with the previously obtained predictions. The second regression model is LASSO, which eliminates irrelevant variables. The modification of original pre-conditioning is proposed in this paper. The usefulness of original pre-conditioning as well as proposed modifications are assessed using simulations. The experiment was carried out for linear model with interactions.

• Details
• Contents