wizata-dsapi 1.3.34__tar.gz → 1.3.35__tar.gz

{wizata_dsapi-1.3.34 → wizata_dsapi-1.3.35}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: wizata_dsapi
-Version: 1.3.34
+Version: 1.3.35
 Summary: Wizata Data Science Toolkit
 Author: Wizata S.A.
 Author-email: info@wizata.com

{wizata_dsapi-1.3.34 → wizata_dsapi-1.3.35}/wizata_dsapi/mlmodel.py

@@ -222,10 +222,6 @@ class ModelInfo:
             obj["alias"] = self.alias
         if self.model_type is not None:
             obj["model_type"] = self.model_type
-        if self.file_format is not None:
-            obj["file_format"] = self.file_format
-        if self.source is not None:
-            obj["source"] = self.source
         if self.property_name is not None:
             obj["property_name"] = self.property_name
         if self.is_active is not None:
@@ -251,7 +247,7 @@ class ModelInfo:
         if "model_type" in obj.keys():
             self.model_type = obj["model_type"]
         if "file_format" in obj.keys():
-            self.key = obj["file_format"]
+            self.file_format = obj["file_format"]
         if "source" in obj.keys():
             self.source = obj["source"]
         if "property_name" in obj.keys():

wizata_dsapi-1.3.35/wizata_dsapi/version.py

@@ -0,0 +1 @@
+__version__ = "1.3.35"

{wizata_dsapi-1.3.34 → wizata_dsapi-1.3.35}/wizata_dsapi/wizata_dsapi_client.py

@@ -1078,55 +1078,38 @@ class WizataDSAPIClient(ApiInterface):
         else:
             raise self.__raise_error(response)
 
-    def register_model(self,
-                       model_key,
-                       train_model,
-                       df: pandas.DataFrame,
-                       scaler=None,
-                       has_anomalies: bool = False,
-                       has_target_feat: bool = False,
-                       experiment_key = None) -> tuple[ModelInfo, pandas.DataFrame]:
-        """
-        Register a Machine Learning model to Wizata.
-        Model is tested by the API against a sample dataframe.
-        :param model_key: logical string id to identify the model.
-        :param train_model: trained model (must be compatible with pickle library)
-        :param df: sample dataframe.
-        :param scaler: scaler (must be compatible with pickle library)
-        :param has_anomalies: True is model generate Anomalies
-        :param has_target_feat: True if model need a target feature to be selected
-        :param experiment_key: Reference of an experiment to which link the generated ML Model
-        :return: registered ML Model , pandas.DataFrame
-        """
-        if model_key is None and isinstance(model_key, str):
-            raise ValueError('Please provide a str Model Key to identify the model.')
-        if train_model is None:
-            raise ValueError('Trained Machine Learning model should not be null')
-        elif df is None:
-            raise ValueError('A sample dataframe must be provided')
-
-        ml_model = ModelInfo(key=model_key)
-        ml_model.trained_model = train_model
-        if scaler is not None:
-            ml_model.scaler = scaler
-        ml_model.has_anomalies = has_anomalies
-        ml_model.has_target_feat = has_target_feat
-        ml_model.input_columns = df.columns
-
-        if experiment_key is not None:
-            ml_model.experimentId = self.get(experiment_key=experiment_key).experiment_id
-
-        try:
-            result_df = predict(df, ml_model)
-            if result_df is not None:
-                ml_model.status = "valid"
-                #TODO: proper method to upload and download from registry to modify this one
-                ## self.upsert(ml_model)
-                return ml_model, result_df
-            else:
-                raise RuntimeError('no dataframe was generated by your model while testing predict capabilities')
-        except Exception as e:
-            raise RuntimeError('not able to validated the model : ' + str(e))
+    def upload_model(self,
+                     model_info: ModelInfo):
+        """
+        upload a model within the model repository.
+        :param model_info: model info, with at least key (+twin, +property, +alias) and trained_model.
+        """
+        if model_info.trained_model is None:
+            raise ValueError("model_info must have a trained model as bytes content pkl or pt")
+        files = {
+            "trained_model": (
+                "trained_model." + model_info.file_format ,
+                model_info.trained_model,
+                "application/octet-stream",
+            )
+        }
+        headers = self.__header()
+        headers.pop("Content-Type", None)
+        response = requests.post(self.__url() + f"models",
+                                 headers=headers,
+                                 data={
+                                    "payload": json.dumps(model_info.to_json())
+                                 },
+                                 files=files)
+        if response.status_code == 200:
+            response_json = response.json()
+            if "identifier" in response_json:
+                key , twin, property, alias = ModelInfo.split_identifier(response_json["identifier"])
+                if alias is not None:
+                    model_info.alias = alias
+            return model_info
+        else:
+            raise self.__raise_error(response)
 
     def upsert_experiment(self, key: str, name: str, pipeline=None, twin=None):
         """

{wizata_dsapi-1.3.34 → wizata_dsapi-1.3.35}/wizata_dsapi.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: wizata-dsapi
-Version: 1.3.34
+Version: 1.3.35
 Summary: Wizata Data Science Toolkit
 Author: Wizata S.A.
 Author-email: info@wizata.com

wizata_dsapi-1.3.34/wizata_dsapi/version.py

@@ -1 +0,0 @@
-__version__ = "1.3.34"