06-comparisons.Rmd

# Comparisons {#comparisons}

** Under Construction **

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = F)
knitr::read_chunk('bin/06-comparisons-code.R')
```

```{r comparisons-setup}
```

This chapter will compare the effectiveness of the methods developped in the previous two chapters, and compare them to more basic approaches to identifying students at-risk.

For example, we know that some CEGEPs have implemented a policy whereby they identify students as being at risk based on their mid-term assessments: if the student receives a certain number of "at-risk" or "failing" results, they are automatically sent an email referring them to academic support services. 

Based on this, we can ask the following research questions : 
 - how effective is this approach at identifying students who drop-out? 
 - how does this approach compare to our models from the previous chapters?

We begin with a basic logistic regression with demographic variables, and as well as the number of each type of results of mid-term assessment, for students in their last term at the college. With these predictors, we try to predict if students are about to graduate, or simply not register again.
```{r model-last-term-status}
```