Currently we're doing this is Facebook's FastText and it works pretty well. Now we need to factor in amount. The reason is,
if two transactions have `COSTCO STORE #5600` in the memo, but one is for $250 and the other is for $50, the first needs to
be categorized as "Office Expenses" (snacks) and the other as "Transportation" (gas).

What is the best approach for making a prediction from 2 variables where one is a regression problem and one is a classification
problem?