We have fun with one-scorching security and also_dummies with the categorical details towards the application studies. To your nan-beliefs, we play with Ycimpute library and you will assume nan beliefs inside mathematical variables . To possess outliers analysis, i pertain Regional Outlier Factor (LOF) for the application analysis. LOF finds and you will surpress outliers research.

Per current loan throughout the application study can have numerous earlier in the day funds. For each and every early in the day software has actually one row which will be acknowledged by the newest feature SK_ID_PREV.

I have each other drift and you can categorical variables. I implement rating_dummies to own categorical parameters and aggregate so you’re able to (indicate, min, maximum, number, and you will share) to possess drift parameters.

The content off percentage history for past finance in the home Borrowing from the bank. There was that line for every made fee plus one row for every missed payment.

With respect to the lost really worth analyses, shed thinking are small. So we don’t need to take one step to own shed values. We have one another float and you will categorical details. We use get_dummies to own categorical details and you will aggregate in order to (indicate, minute, maximum, count, and you will contribution) having drift variables.

This information consists of month-to-month equilibrium pictures out of earlier credit cards one new candidate gotten from your home Borrowing from the bank

It includes month-to-month analysis concerning the earlier in the day credits in the Bureau research. For every row is certainly one day out-of an earlier borrowing, and just one past credit may have several rows, that each month of one’s borrowing size. Read More