Closed felix-zailskas closed 5 months ago
I tried running the demo on my machine, using S3 repository. I passed randomforestclassifier_epochs(20)_f1(0.6259)_model.joblib
as a model file name but got the error Error loading model 'randomforestclassifier_epochs(20)_f1(0.6259)_model.joblib': An error occurred (404) when calling the HeadObject operation: Not Found
When I'm not loading the model from a file, everything works perfectly :)
I tried running the demo on my machine, using S3 repository. I passed
randomforestclassifier_epochs(20)_f1(0.6259)_model.joblib
as a model file name but got the errorError loading model 'randomforestclassifier_epochs(20)_f1(0.6259)_model.joblib': An error occurred (404) when calling the HeadObject operation: Not Found
Could you try loading the .pkl model. There was some shuffling around with file extensions and that one was the one I tried. I will look into it otherwise.
I tried running the demo on my machine, using S3 repository. I passed
randomforestclassifier_epochs(20)_f1(0.6259)_model.joblib
as a model file name but got the errorError loading model 'randomforestclassifier_epochs(20)_f1(0.6259)_model.joblib': An error occurred (404) when calling the HeadObject operation: Not Found
Also could not reproduce this. There was however a compatibility issue with the .joblib models. I moved all incompatible models into the old_models folder on s3 to easily identify them.
New EVP demo has been created. Currently, a Random Forest model can be used to test the way the predictors work now.
To add a new model and test it:
predictors.py::Predictors
Enumpredictors.py::Classifier
class and implement the needed methodsevp.py::EstimatedValuePredictor.__init__
method to allow for initialization of the new predictorThis intermediate code state should make it easy to add new model types, test them, save and load them in a streamlined process so that the best model can easily be used in the end.
Note this does not include final results for the prediction model.