The objective of this project was to build a model that could predict if students were drinking dangerous amounts of alcohol.
The first analysis, student_alcohol_risk, had the students divided into three groups, High, Medium, and Low depending on their drinking habits. The models created did not prove to be overly useful nor statistically significant, although there are some interesting findings in the analysis.
THe second analysis, student_alcohol_risk_two_levels, divided the students into two groups, High and Low. By simplifying the data and the models, the hope was that this would improve the performance of the models. Although the Accuracy and Kappa values increased, these models were still statistically insignificant.
Although I am disappointed that neither of these analysis are overly useful in a practical sense, I was still able to try out some new methods (such as recursive feature elimination), and improve my abilities in R. At the end of the 'two_levels' analysis, I go into some detail about the limitations of the data for building my models.
The easiest way to view my code is to click on the .Rmd files, and if you would like to see the full analysis, please download the .HTML files.
To see more about this dataset, visit: https://www.kaggle.com/uciml/student-alcohol-consumption