Comparison of generalisation error's methods on case of Linear Regression

Abstract

Данный сборник ежегодно проходит индексацию в базе Scopus, в этом году организаторы уже отправили сборник на индексацию.

In feature selection process it is important to estimate generalized error of selected model. There are a lot of methods based on resampling procedure. All this methods have their features, and this work have a large study. Each of selected method was run for 7200 times to nd any pattern according to samples and feature count of dataset. This research shows accuracy of each method for Lasso regression in case of linear function.