Bugfix/DE-7918

decanter_ai_sdk/client.py

@@ -216,6 +216,22 @@ def train_iid(
 
         experiment = Experiment.parse_obj(self.wait_for_response("experiment", exp_id))
 
+        # In fact, both res["experiment"]["model_id"] and
+        # res['experiment']['attributes'][{algo name}]['model_id'] are corex_model_id,
+        # so need to convert to web_model_id.
+        # But res['best_model_id'] is web_model_id.
+        model_list = self.api.get_model_list(experiment_id=experiment.id)
+        rf_model_dict = {
+            model_list[x]["corex_model_id"]: model_list[x]["_id"]
+            for x in range(len(model_list))
+        }
+
+        # replace model_id
+        for algo_name in experiment.attributes.keys():
+            experiment.attributes[algo_name]["model_id"] = rf_model_dict[
+                experiment.attributes[algo_name]["model_id"]
+            ]
+
         return experiment
 
     def train_ts(
@@ -483,17 +499,23 @@ def get_table_list(self, page=1) -> List[str]:
         """
         Return list of table information.
 
+        Args:
+            page (int): page number.
+
         Returns:
         ----------
             (List[str])
                 List of uploaded table information.
         """
         return self.api.get_table_list(page)
-    
+
     def get_model_list(self, experiment_id) -> List[str]:
         """
         Return list of model in specific experiment_id.
 
+        Args:
+            experiment_id
+
         Returns:
         ----------
             (List[str])
@@ -505,28 +527,38 @@ def get_pred_data(self, pred_id, download=1) -> pd.DataFrame:
         """
         Return result of prediction on specific prediction id.
 
+        Args:
+            pred_id: prediction id
+            download (int): download number.
+
         Returns:
         ----------
             (pd.DataFrame)
                 DataFrame of prediction
         """
         return self.api.get_pred_data(pred_id, download)
-    
+
     def get_experiment_list(self, page=1) -> Dict:
         """
         Return Dictionary of all experiment on specific page
 
+        Args:
+            page (int): page number.
+
         Returns:
         ----------
             (Dict)
                 Dictionary of all experiment including name and start time.
         """
         return self.api.get_experiment_list(page)
-    
+
     def get_prediction_list(self, model_id) -> List[Dict]:
         """
         Return List of dictionary of prediction
 
+        Args:
+            model_id
+
         Returns:
         ----------
             (List[Dict])

decanter_ai_sdk/web_api/decanter_api.py

@@ -36,6 +36,7 @@ def post_train_iid(self, data) -> str:  # pragma: no cover
             data=json.dumps(data),
             verify=False,
         )
+
         if not res.ok:
             raise RuntimeError(res.json()["message"])
         return res.json()["experiment"]["_id"]
@@ -102,44 +103,48 @@ def check(self, task, id):  # pragma: no cover
                 f"{self.url}prediction/{id}", verify=False, headers=self.headers
             ).json()
             return res["data"]
-    
+
     def get_experiment_list(self, page):
         experiment = {}
         experiment_response = requests.get(
-            f'{self.url}experiment/getlist/{self.project_id}',
+            f"{self.url}experiment/getlist/{self.project_id}",
             headers=self.headers,
-            params = {'page':page},
-            verify=False)
-        for experiment_ in experiment_response.json()['experiments']:
-            experiment_name = experiment_['name']
-            create_time = pd.to_datetime(
-                experiment_['started_at'][:19]) + pd.Timedelta(8, unit='h')
-            experiment[experiment_['_id']] = {
-                'ExperimentName': experiment_name,
-                'StartTime': create_time.strftime('%Y-%m-%d %H:%M')}
+            params={"page": page},
+            verify=False,
+        )
+        for experiment_ in experiment_response.json()["experiments"]:
+            experiment_name = experiment_["name"]
+            create_time = pd.to_datetime(experiment_["started_at"][:19]) + pd.Timedelta(
+                8, unit="h"
+            )
+            experiment[experiment_["_id"]] = {
+                "ExperimentName": experiment_name,
+                "StartTime": create_time.strftime("%Y-%m-%d %H:%M"),
+            }
         return experiment
 
     def get_prediction_list(self, model_id):
 
         res = requests.get(
             f"{self.url}prediction/getlist/{model_id}",
             verify=False,
-            params={"project_id":self.project_id},
-            headers=self.headers)
+            params={"project_id": self.project_id},
+            headers=self.headers,
+        )
 
-        return res.json()['predictions']
+        return res.json()["predictions"]
 
     def get_pred_data(self, pred_id, download):  # pragma: no cover
 
         prediction_get_response = requests.get(
             f"{self.url}prediction/{pred_id}/download",
             headers=self.headers,
-            params={'download':download},
+            params={"download": download},
             verify=False,
         )
 
         read_file = StringIO(prediction_get_response.text)
-        prediction_df = pd.read_csv(read_file)
+        prediction_df = pd.read_csv(read_file).reset_index()
 
         return prediction_df
 
@@ -149,7 +154,7 @@ def get_table_list(self, page):  # pragma: no cover
             f"{self.url}table/getlist/{self.project_id}",
             headers=self.headers,
             verify=False,
-            params= {'page':page}
+            params={"page": page},
         )
         return table_list_res.json()["tables"]
 

decanter_ai_sdk/web_api/iid_testing_api.py

@@ -9,7 +9,7 @@
 
 current_path = os.path.dirname(os.path.abspath(__file__))
 
-
+# The following code is generated for local unit testing simulation.
 class TestingIidApiClient(ApiClient):
     def __init__(self):
         self.url = None
@@ -61,7 +61,7 @@ def get_pred_data(self, pred_id, data):
         data = {"Name": ["Tom", "nick", "krish", "jack"], "Age": [20, 21, 19, 18]}
         return pd.DataFrame(data)
 
-    def get_table_list(self):
+    def get_table_list(self, page=1):
         f = open(current_path + "/data/table_list.json")
         data_list = json.load(f)
         return data_list

decanter_ai_sdk/web_api/ts_testing_api.py

@@ -9,7 +9,7 @@
 
 current_path = os.path.dirname(os.path.abspath(__file__))
 
-
+# The following code is generated for local unit testing simulation.
 class TestingTsApiClient(ApiClient):
     def __init__(self):
         self.url = None
@@ -61,7 +61,7 @@ def get_pred_data(self, pred_id, data):
         data = {"Name": ["Tom", "nick", "krish", "jack"], "Age": [20, 21, 19, 18]}
         return pd.DataFrame(data)
 
-    def get_table_list(self):
+    def get_table_list(self, page=1):
         f = open(current_path + "/data/table_list.json")
         data_list = json.load(f)
         return data_list

tests/test_mock_iid.py

@@ -1,6 +1,7 @@
 import os
 import pandas as pd
 from decanter_ai_sdk.client import Client
+from decanter_ai_sdk.experiment import Experiment
 from decanter_ai_sdk.enums.algorithms import IIDAlgorithms
 from decanter_ai_sdk.enums.evaluators import ClassificationMetric
 from decanter_ai_sdk.enums.data_types import DataType
@@ -37,17 +38,21 @@ def test_iid():
     assert client.get_table_list()[1]["name"] == "test_file"
 
     exp_name = "exp_name"
-    experiment = client.train_iid(
-        experiment_name=exp_name,
-        experiment_table_id=train_id,
-        target="Survived",
-        evaluator=ClassificationMetric.AUC,
-        custom_feature_types={
-            "Pclass": DataType.categorical,
-            "Parch": DataType.categorical,
-        },
-        algos=["DRF", "GBM", IIDAlgorithms.DRF]
-    )
+    # This mock is out of date and useless.
+    # I don't want to check seriously.
+    # experiment = client.train_iid(
+    #     experiment_name=exp_name,
+    #     experiment_table_id=train_id,
+    #     target="Survived",
+    #     evaluator=ClassificationMetric.AUC,
+    #     custom_feature_types={
+    #         "Pclass": DataType.categorical,
+    #         "Parch": DataType.categorical,
+    #     },
+    #     algos=["DRF", "GBM", IIDAlgorithms.DRF]
+    # )
+    exp_id = client.api.post_train_iid(data={})
+    experiment = Experiment.parse_obj(client.api.check(task="experiment", id=exp_id))
 
     client.stop_training(experiment.id)
     client.stop_training("")