Class 5: content based recommender for MeetFresh

What you should do before the class

Please discuss with your group about how you would like to proceed with content-based recommender using MeetFresh images. Just the plan. You will be doing the experiments during the class. Finish all the exercise of the Shirt Exercise.

Class content

1. Presentation: Present the final results of your Shirt Exercise.
2. Presentation: Present your plan for the MeetFresh use case.
3. Excercise: develop content-based recommender code for MeetFresh.

Note that there are two parts of it:

1. Similarity Measurement
2. Based on users' purchasing history, come up with the recommendation according to content-based item-item similarity.

1. (Optional) Excercise: if you never used PAC in the content-based recommender, try that. Anything suprising about the PCA result? Any improvement to your model?
2. Discussion: How to evaluate the performance of the recommender?
3. (Optional) Exercise: if you still have time, discuss and try to improve your recommender.