TPS 17

Consider the following data set that consists of 10 observations from made up students in a made up class. The first column show the observation id (just for reference; don't use this in your model). The next two columns are features (scores on exams 1 and 2), and the label to target is the final column (the bucketed performance on the final: poor, good, or great). The first table shows the observations sorted by exam1.

The second table is the same data, but sorted by exam2.

Build a decision tree out of this data. You don't have to do anything formal in terms of picking features/thresholds to split on, but you should pick them based on what makes sense by eyeballing the data. Make sure you indicate the class distribution at each leaf.

Spend about 5 minutes on your own thinking through a solution without looking up any material and write it down either in an electronic document (Google Doc, Word, etc.) or on a piece of paper. After the 5 minutes is up, share what you have with your group members. If a revision is necessary, add it as a new section to your document or paper. Upload your document or an image of your paper to this Canvas assignment.

Please keep in mind that you will not be graded based on correctness but on effort. Trust me that it is more helpful in terms of knowledge retention for you to 1) not look at notes while doing the TPS and 2) only look at the solution after you have submitted.

After you have submitted, see this video for solutions.