REFACTOR: Fix gigantic error. Gridsearch returned the worst parameter…

…s, not the best

machine_learning/config.py

@@ -5,10 +5,10 @@ class HPConfig:
  max_leaf_nodes = 500 # værdien fra paperet om modeller er 10
 
 class BestHPConfig:
- criterion = None
- max_depth = None
+ criterion = 'friedman_mse'
+ max_depth = 10
  max_features = None
- max_leaf_nodes = None
+ max_leaf_nodes = 10
 
 class GridSearchConfig:
  param_grid = {

machine_learning/main.py

@@ -27,7 +27,7 @@ def train():
  model = pickle.load(model_file)
  else:
  print("Training...")
- model = train_model(train_data)
+ model = train_model(train_data, grid_search_cv=False)
 
  # Save the trained model
  print("Saving trained model...")

machine_learning/training.py

@@ -3,7 +3,7 @@
 from sklearn.tree import DecisionTreeRegressor
 from machine_learning.prepare_for_training import TrainingDataset
 from sklearn.metrics import mean_squared_error
-from machine_learning.config import HPConfig, GridSearchConfig
+from machine_learning.config import HPConfig, GridSearchConfig, BestHPConfig
 from sklearn.metrics import make_scorer
 
 
@@ -17,7 +17,7 @@ def rmse_cum_power(true_labels, predicted_labels):
 
 
 def custom_scoring(true_labels, predicted_labels):
- return rmse_cum_power(true_labels, predicted_labels)
+ return -rmse_cum_power(true_labels, predicted_labels)
 
 custom_scoring = make_scorer(custom_scoring)
 
@@ -61,8 +61,8 @@ def train_model(train_data, grid_search_cv=True):
  training_dataset = TrainingDataset(train_data)
 
  # Instantiate the decision tree model with specified hyperparameters
- model = DecisionTreeRegressor(criterion=HPConfig.criterion, max_depth=HPConfig.max_depth,
- max_features=HPConfig.max_features, max_leaf_nodes=HPConfig.max_leaf_nodes,
+ model = DecisionTreeRegressor(criterion=BestHPConfig.criterion, max_depth=BestHPConfig.max_depth,
+ max_features=BestHPConfig.max_features, max_leaf_nodes=BestHPConfig.max_leaf_nodes,
  random_state=42)
 
  # Extract features and targets from the training dataset
@@ -124,13 +124,7 @@ def evaluate_model(model, test_data):
  test_targets_cumulative_power = np.cumsum(test_targets_np[:, 0] * test_targets_np[:, 1] * time_diff)
  original_test_rmse = np.sqrt(mean_squared_error(test_targets_cumulative_power, cumulative_power))
 
- print(f"Original Test Root Mean Squared Error (RMSE) for Cumulative Power: {original_test_rmse}")
-
- # Calculate RMSE on the adjusted cumulative power for the test set
- test_rmse = np.sqrt(
- mean_squared_error(test_targets_np[:, 0] * test_targets_np[:, 1], power_consumption_predictions))
- print(f"Adjusted Test Root Mean Squared Error (RMSE) for Cumulative Power: {test_rmse}")
-
+ print(f"Root Mean Squared Error (RMSE) for Cumulative Power: {original_test_rmse}")
  print("Evaluation finished!")
 
  return model