I am working on a project where I have to train a RL agent which will simulate Loan repayment track of a customer's loan based on his features derived from his credit profile (state vector). I am planning to implement episodic SARSA for control, using Tensorforce.

I have gathered some theoretical information for this project but I have not figured out the implementation part yet. My major doubt is that once the agent takes an action I want to update the state vector, I did not find anything on how to go about it. Any help will be appreciated.


  • $\begingroup$ After an action is taken, the new state is given by the environment. Supposing that you are constructing the environment, you need to give the state transitions. As an example, assuming that possible actions are taking out and repaying loans, to decide how the state changes, you need to consider how the credit score changes: if the repayment is delayed, the score can decrease; if it is timely payment the score might increase, etc. Essentially, since you define the environment, you must define how credit profile changes with the actions. $\endgroup$ Apr 28 at 5:27

Your Answer

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

Browse other questions tagged or ask your own question.