A university learning-support office is choosing between TWO…
A university learning-support office is choosing between TWO logistic regression models for flagging students at risk of failing a gateway course. Missing an at-risk student — a student who ends up failing and possibly dropping out — has very high human and financial cost (the university estimates **~$30,000 in lifetime lost tuition per lost student**). Flagging a non-at-risk student costs about **$80 of advisor time** per unneeded outreach. Both models were tested on 1,500 enrolled students; 200 actually failed the course. MODEL P: TN = 1250, FP = 50, FN = 140, TP = 60 MODEL Q: TN = 1100, FP = 200, FN = 30, TP = 170 In 2-3 sentences, recommend which model the office should deploy. Cite specific numbers from the confusion matrices above and connect your recommendation to the relative costs described in the narrative.
Read Details