Fixed minor issues #4
Conversation
| @@ -71,7 +71,7 @@ def choose_action(self, observation): | |||
| # some actions have the same value | |||
| state_action = state_action.reindex(np.random.permutation(state_action.index)) | |||
|
|
|||
| action = state_action.argmax() | |||
| action = state_action.values.argmax() | |||
There was a problem hiding this comment.
This change stops my code from properly selecting an action, what is this change for?
There was a problem hiding this comment.
I was having trouble running your example code -- I haven't spent enough time pinpointing this issue to tell you exactly what's wrong, but I did find a quick fix here.
From what I skimmed it has to do with argmax being deprecated in a newer version of pandas' dtype. The specific error I was getting is TypeError: reduction operation 'argmax' not allowed for this dtype.
I can investigate this further if you think it's worth pursuing, we can simply close the PR as a non-issue, or perhaps the best solution is to include a requirements.txt to lock in the versions of things.
There was a problem hiding this comment.
Hi @ClRobert it turns out you can use action = state_action.idxmax()
No description provided.