Basics of Feature Selection and Statistical Learning for High Energy Physics

TL;DR

This paper introduces fundamental data preparation, feature selection, and statistical learning techniques tailored for high energy physics, including PCA, information gain, classifiers, and toolboxes for application.

Contribution

It presents an overview of feature selection methods and basic classifiers, along with practical toolboxes, tailored specifically for high energy physics data analysis.

Findings

Effective feature selection methods like PCA and information gain are discussed.

Basic classifiers such as MAP and ML are demonstrated with examples.

Toolboxes for applying statistical learning techniques are introduced.

Abstract

This document introduces basics in data preparation, feature selection and learning basics for high energy physics tasks. The emphasis is on feature selection by principal component analysis, information gain and significance measures for features. As examples for basic statistical learning algorithms, the maximum a posteriori and maximum likelihood classifiers are shown. Furthermore, a simple rule based classification as a means for automated cut finding is introduced. Finally two toolboxes for the application of statistical learning techniques are introduced.