0
$\begingroup$

I am working on a personal loan dataset. For each loan, we recorded its credit status monthly after it was drawn by the borrower. Let's say there were 6 status coded by A-F. My project is to use Markov chain model to train the data and estimate the transition matrix as shown below. Then we can predict the future movement of any single loan in probability.

Meanwhile, the dataset contained substantial features for a single loan, like loan amount, borrower age, income, dwelling region, bank account profile, last 90 days bank statement data, some other credit bureau data etc.

I just get this project from the very beginning and all thoughts are rough but not accurate. I need brain storm. Thanks heaps.

How can I use the data properly to predict the next state given the current state? Any assumption should be seriously considered in this project and any advice or experiance for me? I will do it in Python and/or R.

enter image description here

$\endgroup$
0
$\begingroup$

I answered a very similar question on another stack in response to a question about difficulties with an R package. I'm posting the link here - I think it will give you a path forward.

https://stats.stackexchange.com/questions/233851/multi-state-numerical-overflow-using-msm-package-in-r/239988#239988

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.