Summary

This chapter showed how to use multiple linear regression models, one of the most commonly used family of models, to predict numerical and categorical data. Our focus was on showing programming techniques that allow analysts to be more efficient in the projects while keeping their code quality high. We did so by showing how to create different model combinations programatically, measuring the predictive accuracy, and selecting the best one. The techniques used can easily be used with other, more advanced, types of models, and we encourage you to try to improve on the predictive accuracy by using other families of models. In the code that accompanies this book (https://github.com/PacktPublishing/R-Programming-By-Example), you can find an implementation that also uses generalized linear models to produce predictions.

In the following chapter, we will start working with a different and slightly less technical example that uses product data from a hypothetical company to show how to work with manipulative data in a variety of ways and use it with many kinds of visualizations, including 3D, interactive, and geospatial graphs.