NONPARAMETRIC REGRESSION: AN ALTERNATIVE TO THE SCATTER DIAGRAM

Abstract

In this paper is shown the possibility to use a nonparametric regression instead of a scatter diagram, when this last cannot be employed in order to obtain information about the shape of the model. In particular it is considered a logistic regression, and a single quantitative independent variable. The proposal is illustrated with two examples. The former considers a scatter diagram that shows a linear relationship between the values of the independent variable and the probabilities of success of an event. The other one presents a scatter diagram that evidences a nonlinear relation between them.