Case study - statistical analysis, and build a classifier for predicting high cost patients. This is timed to be performed in 3 hours. Totally ridiculous dataset - you'll get 18 files to analyze with each file containing from 120 - 20k rows, and each file has about 10-35+ columns. And you won't get any data schema, so you'll have no idea on the PK, FK. And PK are different for each file. You'll run out of memory, so analyze each file separately. And also look into the from_date from claims transactions file to include only latest transactions, as some of it is from 1929.
