ABSTRACT
Statistical Modeling is a way to approximate the mechanisms or the rules that govern the functioning of phenomena. The main approaches to statistical modeling are the Data Modeling Culture and the Algorithmic Modeling Culture, corresponding to the traditional and the modern concept of modeling. The Algorithmic Modeling Culture takes a different approach and shifts focus from mathematical models to the properties of algorithms. Instead of developing some elegantly designed model, algorithms try to recreate the black box mechanism by automatically and iteratively finding the way to adapt their output to data. This chapter addresses linear models, the most traditional statistical models in the Data Modeling Culture, and two nonparametric regression techniques. Although the main part of modeling methods is designed to explain multivariate relationships, the chapter focuses on the case of only one explanatory variable and exploits the R functions. Three types of simple regression analysis, expectedpts and scoringprob are designed to deal with analytics needs with a nonparametric approach.
