You only need to do questions 1 and 2 in homework five if you’ve downloaded the homework before October 22. If you’ve downloaded the homework as of 10/22, then complete the entirety of the homework, following the instructions in the homework. These are Bayes Rules and the Conditional Probability proof. These questions are the most relevant to the midterm exam.

Midterm Preparation

The midterm exam will consist of ten questions, and students will have the entire three hours and twenty minutes to complete it. All formulas will be provided; no cheatsheets, computers, or electronic devices are allowed. (You can bring your phones, just don’t use them during the exam.) We will provide scratch paper. The breakdown of these questions are:

  • Map Reduce Questions (30%)
  • Principle Component Analysis (10%)
  • Maximum Likelihood Questions (10%)
  • Unsupervised Machine Learning Questions (10%)
  • Multi-Part Frequent Itemsets and Association Rules Questions (20%)
  • Fresh Content Bayesian Rules Questions (20%)

The midterm is close book without any notes. I will provide some information on the board so that memorization of formulas is unnecessary. Among the information that I anticipate as helpful, here are some that might help:

  • Commonly used map-reduce functions: map, flatMap, reduceByKey, and filter
  • Some helpful derivatives and math properties: log, exponent, product
  • Probabilistic rules: conditional probability, total probability, and Bayes rule

I would suggest that you practice these so that you needn’t look up at the board so frequently.

Midterm Review Slides

Here are some practice problems that should help you prepare: