Basic recommendation engine

[vishalbollu]

Closes #58

---

Checklist:

• Run make test and make lint
• Test end to end manually (e.g. build/push all images, restart local operator, and run cx refresh in an example folder)
• Update documentation
• Update examples and cx init
• Alert team if dev environment changed
• Cherry-pick into release branches if it's a bugfix
• Delete the branch once it's merged

[vishalbollu]

• showcases a collaborative filtering approach to recommendation system using deep learning
• uses movielens 100k dataset dataset as input
• dataset is a zip file containing ratings.csv, movies.csv and other information
• our pipeline will use ratings.csv to group similar users to each other and similar movies to eachother and predict the rating that a user may give to a movie
• dataset contains the reviews of 600 users for 9000 movies with the schema: userid,movieid,rating,timestamp
• the userid range is [1, 610] and movieid range from [1, 193609]
• movieid range is large compared to the number of unique movies in dataset so shrink the range by creating a new index of only the movies that are present in the dataset
• index userid and movieid columns by treating them as string and using index_string
• embed indexed userid and movieid columns
• train a regression model to minimize RMSE for rating predictions
• predict the rating of a user with a userid that is mentioned in the input dataset to an unrated movie with movieid that is in the input dataset

[deliahu]

Can you split embedding_size for each of the features?