A study in alcohol : A comparison of data mining methods for identifying binge drinking risk factors in university students

Sammanfattning: Hazardous alcohol consumption is an issue that affects a lot of university students today. Consuming alcohol tend to have a negative impact on both mental and physical aspects, which can lead to severe alcohol addictions in the future. This study investigates which background factors that causes the phenomenon of binge drinking by collecting and analysing data from Linköping University. The results were analysed with data mining techniques such as: decision trees, random forest, and logistic regression. The results showed that logistic regression were the most reliable method in predicting binge drinking with an accuracy of (86.50 %), precision (92.64 %) and recall (90.96 %). The findings also showed that participation in student events together with higher weekly alcohol consumption predicted binge drinking. Additionally, other risk factors were the amounts of time the students spent with their friends and the students activity in partaking in their programs section (program association). The results from this study suggest that the student culture not only influence alcohol consumption but it induces the habits of binge drinking.