@ParulSinha wrote:
I was trying to solve the ‘German Credit Risk classification’ which aims at predicting if a customer has a good credit or a bad credit
The dataset has only 1000 rows and around 9 variable.
I tried almost all models and though CART and RF work better than other models , I am unable to push F1 score beyond 0.437 …
Tried adding feature(credit.amount/duration),add loss param in CART, log transformations and while all this does show improvement ,its not really impressive.Will this score be too less for such a small data set?
Target Variable is credit.rating
Predictors : Age,
Savings.account(little,moderate,quite rich,rich)
Checking.accounts(little,moderate,quite rich,rich)
Job(0-unskilled,non resident,1 unskilled resident,2 -skilled resident,3-highly skilled)
Credit.amount
Duration
Housing(own,rent,free)
Purpose
Can someone please provide guidanceRegards,
Parul
Posts: 1
Participants: 1