8 multiple choice questions and 1 coding question. The former covered neural networks (how to avoid overfitting training data), decision trees (what are they, and choose the best one given 4 choices), logistic regression (analyze a classification problem where precision and recall are good for one label but not for the other), k-means ("all of the following are true except..."), SVM optimization (how will increasing the weight of the slack variable term in the loss function affect things), SVM loss function (what it looks like), variance in probability (basically summing random variables with constants out front, and finding the overall variance). Coding question involved reading files and writing new ones.