black format all the code (#16)

* Fixed PR changes for params_regression

* Fixed Unittest for param_regression according to PR change

* Fixed the mobility notebook and radp_library file(PR changes done)

* Resolved test_param_regression unittest cases

* Updated Param Regression and radp library as instructed in the PR review

* Changed the scipy dependency to solve dependency error

* Changed how seed value works especially made it user friendly

* Refactored functions and changed the imports

* Resolved test_param_regression imports

* All Black Changes

apps/coverage_capacity_optimization/cco_engine.py

@@ -39,7 +39,6 @@ def rf_to_coverage_dataframe(
         over_coverage_threshold: float = 0,
         growth_rate: float = 1,
     ) -> pd.DataFrame:
-
         if lambda_ <= 0 or lambda_ >= 1:
             raise ValueError("lambda_ must be between 0 and 1 (noninclusive)")
 
@@ -65,13 +64,20 @@ def rf_to_coverage_dataframe(
         coverage_dataframe["weak_coverage"] = np.minimum(0, h)
         coverage_dataframe["overly_covered"] = (h > 0) & (g <= 0)
         coverage_dataframe["over_coverage"] = np.minimum(0, g)
-        coverage_dataframe["covered"] = ~coverage_dataframe["weakly_covered"] & ~coverage_dataframe["overly_covered"]
+        coverage_dataframe["covered"] = (
+            ~coverage_dataframe["weakly_covered"]
+            & ~coverage_dataframe["overly_covered"]
+        )
 
         # TODO : deprecate the below notion
         # soft_weak_coverage = sigmoid(h, growth_rate)
         # soft_over_coverage = sigmoid(g, growth_rate)
-        coverage_dataframe["soft_weak_coverage"] = 1000 * np.tanh(0.05 * growth_rate * h)
-        coverage_dataframe["soft_over_coverage"] = 1000 * np.tanh(0.05 * growth_rate * g)
+        coverage_dataframe["soft_weak_coverage"] = 1000 * np.tanh(
+            0.05 * growth_rate * h
+        )
+        coverage_dataframe["soft_over_coverage"] = 1000 * np.tanh(
+            0.05 * growth_rate * g
+        )
         coverage_dataframe["network_coverage_utility"] = (
             lambda_ * coverage_dataframe["soft_weak_coverage"]
             + (1 - lambda_) * coverage_dataframe["soft_over_coverage"]
@@ -83,8 +89,12 @@ def get_weak_over_coverage_percentages(
         coverage_dataframe: pd.DataFrame,
     ) -> Tuple[float, float]:
         n_points = len(coverage_dataframe.index)
-        weak_coverage_percent = 100 * coverage_dataframe["weakly_covered"].sum() / n_points
-        over_coverage_percent = 100 * coverage_dataframe["overly_covered"].sum() / n_points
+        weak_coverage_percent = (
+            100 * coverage_dataframe["weakly_covered"].sum() / n_points
+        )
+        over_coverage_percent = (
+            100 * coverage_dataframe["overly_covered"].sum() / n_points
+        )
         return weak_coverage_percent, over_coverage_percent
 
     @staticmethod
@@ -122,18 +132,26 @@ def get_cco_objective_value(
                     coverage_dataframe,
                 )
             )
-            augmented_coverage_df_with_normalized_traffic_model["network_coverage_utility"] = (
-                augmented_coverage_df_with_normalized_traffic_model["normalized_traffic_statistic"]
+            augmented_coverage_df_with_normalized_traffic_model[
+                "network_coverage_utility"
+            ] = (
+                augmented_coverage_df_with_normalized_traffic_model[
+                    "normalized_traffic_statistic"
+                ]
                 * coverage_dataframe["network_coverage_utility"]
             )
-            coverage_dataframe["network_coverage_utility"] = augmented_coverage_df_with_normalized_traffic_model[
+            coverage_dataframe[
+                "network_coverage_utility"
+            ] = augmented_coverage_df_with_normalized_traffic_model[
                 "network_coverage_utility"
             ]
 
         if active_ids_list is None:
             return -math.inf
 
-        active_df = coverage_dataframe[coverage_dataframe[id_field].isin(active_ids_list)]
+        active_df = coverage_dataframe[
+            coverage_dataframe[id_field].isin(active_ids_list)
+        ]
         active_sector_metric = active_df.groupby(id_field)["network_coverage_utility"]
 
         if cco_metric == CcoMetric.PIXEL:
@@ -161,7 +179,9 @@ def add_tile_x_and_tile_y(
             Dataframe with tile_x and tile_y columns appended
 
         """
-        tile_coords = list(zip(coverage_dataframe[loc_x_field], coverage_dataframe[loc_y_field]))
+        tile_coords = list(
+            zip(coverage_dataframe[loc_x_field], coverage_dataframe[loc_y_field])
+        )
 
         coverage_dataframe["tile_x"], coverage_dataframe["tile_y"] = zip(
             *map(
@@ -200,11 +220,16 @@ def augment_coverage_df_with_normalized_traffic_model(
                 "over_coverage",
         """
 
-        sum_of_desired_traffic_statistic_across_all_tiles = traffic_model_df[desired_traffic_statistic_col].sum()
+        sum_of_desired_traffic_statistic_across_all_tiles = traffic_model_df[
+            desired_traffic_statistic_col
+        ].sum()
         traffic_model_df["normalized_traffic_statistic"] = (
-            traffic_model_df[desired_traffic_statistic_col] / sum_of_desired_traffic_statistic_across_all_tiles
+            traffic_model_df[desired_traffic_statistic_col]
+            / sum_of_desired_traffic_statistic_across_all_tiles
+        )
+        coverage_dataframe_with_bing_tiles = CcoEngine.add_tile_x_and_tile_y(
+            coverage_df
         )
-        coverage_dataframe_with_bing_tiles = CcoEngine.add_tile_x_and_tile_y(coverage_df)
         augmented_coverage_df_with_normalized_traffic_model = pd.merge(
             traffic_model_df,
             coverage_dataframe_with_bing_tiles,
@@ -235,6 +260,8 @@ def traffic_normalized_cco_metric(coverage_dataframe: pd.DataFrame) -> float:
         # only one of weak_coverage and over_coverage can be simultaneously 1
         # so, the logic below does not double count
         return (
-            coverage_dataframe["normalized_traffic_statistic"] * coverage_dataframe["weak_coverage"]
-            + coverage_dataframe["normalized_traffic_statistic"] * coverage_dataframe["over_coverage"]
+            coverage_dataframe["normalized_traffic_statistic"]
+            * coverage_dataframe["weak_coverage"]
+            + coverage_dataframe["normalized_traffic_statistic"]
+            * coverage_dataframe["over_coverage"]
         ).sum()

apps/coverage_capacity_optimization/cco_example_app.py

@@ -52,7 +52,9 @@
 )
 
 # resolve the model status -- this blocking call ensures training is done and model is available for use
-model_status: ModelStatus = radp_helper.resolve_model_status(MODEL_ID, wait_interval=3, max_attempts=10, verbose=True)
+model_status: ModelStatus = radp_helper.resolve_model_status(
+    MODEL_ID, wait_interval=3, max_attempts=10, verbose=True
+)
 
 # handle an exception if one occurred
 if not model_status.success:

apps/coverage_capacity_optimization/dgpco_cco.py

@@ -137,7 +137,11 @@ def _single_step(
             """Single step of DGPCO."""
 
             # calculate new metric
-            (current_rf_dataframe, current_coverage_dataframe, current_cco_objective,) = self._calc_metric(
+            (
+                current_rf_dataframe,
+                current_coverage_dataframe,
+                current_cco_objective,
+            ) = self._calc_metric(
                 lambda_=lambda_,
                 weak_coverage_threshold=weak_coverage_threshold,
                 over_coverage_threshold=over_coverage_threshold,
@@ -151,8 +155,12 @@ def _single_step(
             # pull the cell config index
             cell_config_index = self.config.index[self.config["cell_id"] == cell_id][0]
 
-            orig_el_idx = self.valid_configuration_values[constants.CELL_EL_DEG].index(orig_el_deg)
-            cur_el_idx = self.valid_configuration_values[constants.CELL_EL_DEG].index(cur_el_deg)
+            orig_el_idx = self.valid_configuration_values[constants.CELL_EL_DEG].index(
+                orig_el_deg
+            )
+            cur_el_idx = self.valid_configuration_values[constants.CELL_EL_DEG].index(
+                cur_el_deg
+            )
 
             for d in opt_delta:
                 new_el_idx = orig_el_idx + d
@@ -161,11 +169,15 @@ def _single_step(
                     # we do not want to check current value
                     continue
 
-                if new_el_idx < 0 or new_el_idx >= len(self.valid_configuration_values[constants.CELL_EL_DEG]):
+                if new_el_idx < 0 or new_el_idx >= len(
+                    self.valid_configuration_values[constants.CELL_EL_DEG]
+                ):
                     # we do not want to wrap around, since that would not be a neighboring tilt
                     continue
 
-                new_el = self.valid_configuration_values[constants.CELL_EL_DEG][new_el_idx]
+                new_el = self.valid_configuration_values[constants.CELL_EL_DEG][
+                    new_el_idx
+                ]
 
                 # update the cell config el_degree
                 self.config.loc[cell_config_index, constants.CELL_EL_DEG] = new_el
@@ -258,7 +270,12 @@ def _single_step(
             logging.info(f"\nIn epoch: {epoch:02}/{num_epochs}...")
 
             # Perform one step of DGPCO
-            (new_opt_el, new_rf_dataframe, new_coverage_dataframe, new_cco_objective_value,) = _single_step(
+            (
+                new_opt_el,
+                new_rf_dataframe,
+                new_coverage_dataframe,
+                new_cco_objective_value,
+            ) = _single_step(
                 cell_id=cell_id,
                 orig_el_deg=orig_el_deg,
                 cur_el_deg=cur_el_deg,

apps/coverage_capacity_optimization/tests/test_cco_engine.py

@@ -14,7 +14,9 @@
 class TestCCO(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
-        cls.dummy_df = pd.DataFrame(data={CELL_ID: [1, 2, 73], LOC_X: [3, 4, 89], LOC_Y: [7, 8, 10]})
+        cls.dummy_df = pd.DataFrame(
+            data={CELL_ID: [1, 2, 73], LOC_X: [3, 4, 89], LOC_Y: [7, 8, 10]}
+        )
 
     def test_invalid_lambda(self):
         self.dummy_df["rsrp_dbm"] = [98, 92, 86]
@@ -37,7 +39,8 @@ def testing_weakly_covered(self):
             self.dummy_df, weak_coverage_threshold=-100, over_coverage_threshold=0
         )
         self.assertEqual(
-            returned_df["weakly_covered"][returned_df["weakly_covered"] == 1].count() == 1,
+            returned_df["weakly_covered"][returned_df["weakly_covered"] == 1].count()
+            == 1,
             True,
         )
 
@@ -58,7 +61,8 @@ def test_overly_covered(self):
             self.dummy_df, weak_coverage_threshold=-100, over_coverage_threshold=0
         )
         self.assertEqual(
-            returned_df["overly_covered"][returned_df["overly_covered"] == 0].count() == 1,
+            returned_df["overly_covered"][returned_df["overly_covered"] == 0].count()
+            == 1,
             True,
         )
 
@@ -81,11 +85,13 @@ def testing_some_not_weakly_or_overcovered(self):
             True,
         )
         self.assertEqual(
-            returned_df["weakly_covered"][returned_df["weakly_covered"] == 1].count() == 1,
+            returned_df["weakly_covered"][returned_df["weakly_covered"] == 1].count()
+            == 1,
             True,
         )
         self.assertEqual(
-            returned_df["overly_covered"][returned_df["overly_covered"] == 1].count() == 1,
+            returned_df["overly_covered"][returned_df["overly_covered"] == 1].count()
+            == 1,
             True,
         )
 
@@ -168,11 +174,14 @@ def test_get_cco_objective_value(self):
         )
         # asserting the multiplied version of network_coverage_utility
         self.assertTrue(
-            coverage_df["network_coverage_utility"][0] == 0.24 and coverage_df["network_coverage_utility"][1] == 0.24
+            coverage_df["network_coverage_utility"][0] == 0.24
+            and coverage_df["network_coverage_utility"][1] == 0.24
         )
 
         # asserting the average of network_coverage_utility
-        self.assertTrue(0.24 == coverage_df["network_coverage_utility"].sum() / len(coverage_df))
+        self.assertTrue(
+            0.24 == coverage_df["network_coverage_utility"].sum() / len(coverage_df)
+        )
 
         # asserting the returned cco_objective_value
         expected_value = 0.24

apps/energy_savings/energy_savings_gym.py

@@ -98,7 +98,9 @@ def __init__(
         self.prediction_dfs = dict()
         for cell_id in site_config_df.cell_id:
             prediction_dfs = BayesianDigitalTwin.create_prediction_frames(
-                site_config_df=self.site_config_df[self.site_config_df.cell_id.isin([cell_id])].reset_index(),
+                site_config_df=self.site_config_df[
+                    self.site_config_df.cell_id.isin([cell_id])
+                ].reset_index(),
                 prediction_frame_template=prediction_frame_template[cell_id],
             )
             self.prediction_dfs.update(prediction_dfs)
@@ -140,11 +142,13 @@ def __init__(
 
         # Reward when all cells are off:
         self.r_norm = (1 - lambda_) * (
-            -10 * np.log10(self.num_cells) - over_coverage_threshold + min_rsrp - weak_coverage_threshold
+            -10 * np.log10(self.num_cells)
+            - over_coverage_threshold
+            + min_rsrp
+            - weak_coverage_threshold
         )
 
     def _next_observation(self):
-
         if self.ue_tracks:
             data = next(self.ue_tracks)
             for batch in data:
@@ -198,25 +202,32 @@ def _next_observation(self):
         )
         if self.traffic_model_df is None:
             cco_objective_metric = (
-                coverage_dataframe["weak_coverage"].mean() + coverage_dataframe["over_coverage"].mean()
+                coverage_dataframe["weak_coverage"].mean()
+                + coverage_dataframe["over_coverage"].mean()
             )
 
         else:
-            processed_coverage_dataframe = CcoEngine.augment_coverage_df_with_normalized_traffic_model(
-                self.traffic_model_df,
-                "avg_of_average_egress_kbps_across_all_time",
-                coverage_dataframe,
+            processed_coverage_dataframe = (
+                CcoEngine.augment_coverage_df_with_normalized_traffic_model(
+                    self.traffic_model_df,
+                    "avg_of_average_egress_kbps_across_all_time",
+                    coverage_dataframe,
+                )
             )
 
-            cco_objective_metric = CcoEngine.traffic_normalized_cco_metric(processed_coverage_dataframe)
+            cco_objective_metric = CcoEngine.traffic_normalized_cco_metric(
+                processed_coverage_dataframe
+            )
 
         # Output for debugging/postprocessing purposes
         if self.debug:
             self.rf_dataframe = rf_dataframe
             self.coverage_dataframe = coverage_dataframe
 
         return (
-            EnergySavingsGym.ENERGY_MAX_PER_CELL * sum(self.on_off_state) / len(self.on_off_state),
+            EnergySavingsGym.ENERGY_MAX_PER_CELL
+            * sum(self.on_off_state)
+            / len(self.on_off_state),
             0.0,
             cco_objective_metric,  # TODO : normalized this against MAX_CLUSTER_CCO
         )
@@ -230,7 +241,11 @@ def reward(
         if energy_consumption == 0:
             return self.r_norm
         else:
-            return self.lambda_ * -1.0 * energy_consumption + (1 - self.lambda_) * cco_objective_metric - self.r_norm
+            return (
+                self.lambda_ * -1.0 * energy_consumption
+                + (1 - self.lambda_) * cco_objective_metric
+                - self.r_norm
+            )
 
     def make_action_from_state(self):
         action = np.empty(self.num_cells, dtype=int)
@@ -242,7 +257,6 @@ def make_action_from_state(self):
         return action
 
     def _take_action(self, action):
-
         num_cells = len(self.site_config_df)
         # on_off_cell_state captures the on/off state of each cell (on is `1`)
         on_off_cell_state = [1] * num_cells
@@ -277,7 +291,6 @@ def reset(self):
         return self._next_observation()
 
     def step(self, action):
-
         # Execute one time step within the environment
         self._take_action(action)
 
@@ -293,7 +306,9 @@ def step(self, action):
 
         return obs, reward, done, {}
 
-    def get_all_possible_actions(self, possible_actions: List[List[int]]) -> List[List[int]]:
+    def get_all_possible_actions(
+        self, possible_actions: List[List[int]]
+    ) -> List[List[int]]:
         """
         A recursive function to get all possible actions as a list.
         Useful for bruteforce search.

apps/example/example_app.py

@@ -136,9 +136,7 @@
     train_response = radp_client.train(
         model_id=MODEL_ID,
         params=TRAINING_PARAMS,
-        ue_training_data=pd.concat(
-            [pd.read_csv(file) for file in TRAINING_DATA_FILES]
-        ),
+        ue_training_data=pd.concat([pd.read_csv(file) for file in TRAINING_DATA_FILES]),
         topology=pd.read_csv(TOPOLOGY_FILE),
     )
 
@@ -164,9 +162,7 @@
 # run simulation on cumulative data passed to model
 simulation_response = radp_client.simulation(
     simulation_event=simulation_event,
-    ue_data=pd.concat(
-        [pd.read_csv(file) for file in PREDICTION_DATA_FILES]
-    ),
+    ue_data=pd.concat([pd.read_csv(file) for file in PREDICTION_DATA_FILES]),
     config=pd.read_csv(PREDICTION_CONFIG),
 )
 simulation_id = simulation_response["simulation_id"]