Hello,
I'm trying to understand the principles behind the models for homework 3. Can someone give me a plain-English explanation for why the addition of the mid-semester testing score (G2) to otherwise largely demographic data, the linear model changes the model's predictors and improves accuracy so much. However, when we add the G2 variable in a similar maneuver to the support vector machine model, it doesn't impact the accuracy much (unless I did something wrong?) and doesn't demonstrate the same compelling predictive value?
I assume this reflects the underlying statistical principles of the modeling. Can someone explain this?
Thanks
Julia