I am relatively new to reinforcement learning and have been experiencing with a reinforcement learning model to recommend actions to patients. Appreciate if someone can help me in understanding how best to evaluate a reinforcement learning model for performance before the model goes into production and when it is in production.
I’m not pro either but I would say you should print the reward that your code is making. From reward you should see is the code better than other. For example if you want to make rl algorithm which sell and buy stocks then the “accuracy” is how much the model make profit. You can use old data to evaluate it. Then you can also try to find some baselines if the problem is common. I’m not sure was this your question. If it wasn’t ask it different way and I can try to help you.
Can you share how the model is being trained?