added model tunning into dvc pipeline and execute it

dvc.lock

@@ -94,8 +94,8 @@ stages:
       size: 1932798
     - path: ./src/models/train_model.py
       hash: md5
-      md5: 19e77eb2eec71235661cbea2f088c3d6
-      size: 16692
+      md5: 90f6f48a40e04b569de4dfe36fc48c71
+      size: 16665
     params:
       params.yaml:
         base.target: Time_taken(min)
@@ -145,3 +145,31 @@ stages:
       hash: md5
       md5: 4804de180ead9bca13523f33d5e725b0
       size: 866646
+  tune_model:
+    cmd: python ./src/models/tune_model.py
+    deps:
+    - path: ./data/processed/processed_test.csv
+      hash: md5
+      md5: 8fbf4dba54f10d8f1c9d6b1c85b81041
+      size: 435216
+    - path: ./data/processed/processed_train.csv
+      hash: md5
+      md5: ae10fc577ecbf3cfe7311c1af6e59e2c
+      size: 1932798
+    - path: ./src/models/tune_model.py
+      hash: md5
+      md5: 9f7e26afd8ac368de8c0de728cdd0fb2
+      size: 7639
+    params:
+      params.yaml:
+        base.target: Time_taken(min)
+        feature_engineering.export_path: /data/processed
+        mlflow.repo_name: delivery-time-estm
+        mlflow.repo_owner: ronylpatil
+        tune_model.model_name: xgb_tunned
+        tune_model.n_trials: 50
+    outs:
+    - path: ./models/xgb_tunned.joblib
+      hash: md5
+      md5: e45b2fd0bfcb258d1b3fa27cfdc77132
+      size: 253468

dvc.yaml

@@ -63,3 +63,19 @@ stages:
     - train_model.xgb
     outs:
     - ./models/${train_model.model_name}.joblib
+
+  tune_model:
+    cmd: python ./src/models/tune_model.py
+    deps:
+    - ./src/models/tune_model.py
+    - .${feature_engineering.export_path}/${feature_engineering.filename_train}.csv
+    - .${feature_engineering.export_path}/${feature_engineering.filename_test}.csv
+    params:
+    - base.target
+    - feature_engineering.export_path
+    - mlflow.repo_name
+    - mlflow.repo_owner
+    - tune_model.model_name
+    - tune_model.n_trials
+    outs:
+    - ./models/${tune_model.model_name}.joblib

models/.gitignore

@@ -1 +1,2 @@
 /model_rf.joblib
+/xgb_tunned.joblib

params.yaml

@@ -72,3 +72,7 @@ train_model:
     max_depth: 7 # {}
     lambda: 1
     alpha: 0
+
+tune_model:
+  model_name: xgb_tunned
+  n_trials: 50

src/models/tune_model.py

@@ -81,7 +81,7 @@ def objective_xgb(trial) -> float:
 
     # Perform 3-fold cross-validation and calculate accuracy
     score = cross_val_score(
-        model, x_train, y_train, cv=5, scoring="neg_mean_squared_error"
+        model, x_train, y_train, cv=5, scoring="neg_mean_absolute_error"
     ).mean()
 
     return score  # Return the accuracy score for Optuna to maximize
@@ -182,7 +182,7 @@ def objective_xgb(trial) -> float:
         mlflow.sklearn.log_model(best_model, "tunned xgbR", signature=signature)
         mlflow.set_tag("developer", "ronil")
         mlflow.set_tag("model", "xgbR")
-        mlflow.set_tag("objective", "neg_mean_squared_error")
+        mlflow.set_tag("objective", "neg_mean_absolute_error")
         infologger.info("Experiment tracked successfully.")
 
         save_model(