Section 15 Prediction and Classification

This section will provide a brief introduction to tools related to the goals of prediction and classification.

In the social sciences, we often have one of a few goals in mind:

• Describe
• Explain, evaluate, and recommend \(\rightarrow\) Causality
• Discover and Predict

Most of the tools we have been working on thus far have focused on first describing our data and then conducting hypothesis tests through different types of regression, in order to assess a deductive hypothesis, explaining the relationship between two variables.

Here, we are going to turn to the third goal above.

Some supplemental resources for this section include the following R packages:

• I highly recommend this online interactive course (free- amazing!) from Jilia Silge which walks through a newer package, tidymodels similar to caret, that can apply several different types of machine learning models. It draws on the fundamentals covered in this section but adds more advanced techniques. Here is an additional detailed tutorial on this package here and here.
• caret package and presentation
• RTextTools (may or may not be updated) here and here