From 3c368e96e9035c5f4b1d3c03e40a7dbd16c04d67 Mon Sep 17 00:00:00 2001
From: rahel_yab <rahelyabebal@gmail.com>
Date: Tue, 16 Dec 2025 16:40:22 +0300
Subject: [PATCH 1/3] added model files to .gitignore

---
 .gitignore | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/.gitignore b/.gitignore
index 62efb6d..7fe5334 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,9 @@
 /data
 /venv
 */__pycache__
+/models
+
+models/
+*.pkl
+*.joblib
+mlflow.db
\ No newline at end of file

From e6ef1d91fd4032ffae2f394532ac704e05691e89 Mon Sep 17 00:00:00 2001
From: rahel_yab <rahelyabebal@gmail.com>
Date: Tue, 16 Dec 2025 16:40:53 +0300
Subject: [PATCH 2/3] added fastapi, mlflow, joblib to requirements

---
 requirements.txt | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/requirements.txt b/requirements.txt
index 974177c..a9cfd15 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -3,4 +3,9 @@ numpy
 scipy
 matplotlib
 seaborn
-pytest
\ No newline at end of file
+pytest
+joblib
+mlflow
+fastapi
+uvicorn
+flake8
\ No newline at end of file

From 78b985a52cc9d04afb16da0c03a7f4ca4e8cf6ce Mon Sep 17 00:00:00 2001
From: rahel_yab <rahelyabebal@gmail.com>
Date: Tue, 16 Dec 2025 16:41:14 +0300
Subject: [PATCH 3/3] optimized training using mlflow

---
 src/train.py | 69 +++++++++++++++++++++++++++++++++++++---------------
 1 file changed, 49 insertions(+), 20 deletions(-)

diff --git a/src/train.py b/src/train.py
index 72663ba..318d07f 100644
--- a/src/train.py
+++ b/src/train.py
@@ -1,6 +1,8 @@
 import pandas as pd
 import joblib
 from pathlib import Path
+import mlflow
+import mlflow.sklearn
 
 from sklearn.model_selection import train_test_split
 from sklearn.pipeline import Pipeline
@@ -62,33 +64,60 @@ def build_pipeline(X: pd.DataFrame) -> Pipeline:
     )
 
 def main():
-    print("Loading data...")
-    df = load_data(DATA_PATH)
+    mlflow.set_experiment("Credit Risk Model")
 
-    print("Preparing features and target...")
-    X = df.drop(columns=[TARGET_COL] + DROP_COLS, errors="ignore")
-    y = df[TARGET_COL]
+    with mlflow.start_run():
+        print("Loading data...")
+        df = load_data(DATA_PATH)
 
-    X_train, X_test, y_train, y_test = train_test_split(
-        X, y, test_size=0.2, random_state=42, stratify=y
-    )
+        print("Preparing features and target...")
+        X = df.drop(columns=[TARGET_COL] + DROP_COLS, errors="ignore")
+        y = df[TARGET_COL]
+
+        X_train, X_test, y_train, y_test = train_test_split(
+            X, y, test_size=0.2, random_state=42, stratify=y
+        )
+
+        print("Building model pipeline...")
+        pipeline = build_pipeline(X_train)
+
+        # Log parameters
+        mlflow.log_param("model_type", "LogisticRegression")
+        mlflow.log_param("max_iter", 1000)
+        mlflow.log_param("class_weight", "balanced")
+        mlflow.log_param("random_state", 42)
+        mlflow.log_param("test_size", 0.2)
+
+        print("Training model...")
+        pipeline.fit(X_train, y_train)
+
+        print("Evaluating model...")
+        y_pred = pipeline.predict(X_test)
+        y_proba = pipeline.predict_proba(X_test)[:, 1]
+
+        report = classification_report(y_test, y_pred, output_dict=True)
+        roc_auc = roc_auc_score(y_test, y_proba)
 
-    print("Building model pipeline...")
-    pipeline = build_pipeline(X_train)
+        print(classification_report(y_test, y_pred))
+        print("ROC-AUC:", roc_auc)
 
-    print("Training model...")
-    pipeline.fit(X_train, y_train)
+        # Log metrics
+        mlflow.log_metric("accuracy", report["accuracy"])
+        mlflow.log_metric("precision", report["weighted avg"]["precision"])
+        mlflow.log_metric("recall", report["weighted avg"]["recall"])
+        mlflow.log_metric("f1_score", report["weighted avg"]["f1-score"])
+        mlflow.log_metric("roc_auc", roc_auc)
 
-    print("Evaluating model...")
-    y_pred = pipeline.predict(X_test)
-    y_proba = pipeline.predict_proba(X_test)[:, 1]
+        # Log model
+        mlflow.sklearn.log_model(pipeline, "model")
 
-    print(classification_report(y_test, y_pred))
-    print("ROC-AUC:", roc_auc_score(y_test, y_proba))
+        model_file = MODEL_PATH / "credit_risk_logistic.pkl"
+        joblib.dump(pipeline, model_file)
+        print(f"Model saved to {model_file}")
 
-    model_file = MODEL_PATH / "credit_risk_logistic.pkl"
-    joblib.dump(pipeline, model_file)
-    print(f"Model saved to {model_file}")
+        # Register the model
+        model_uri = f"runs:/{mlflow.active_run().info.run_id}/model"
+        mlflow.register_model(model_uri, "CreditRiskLogisticRegression")
 
 if __name__ == "__main__":
     main()