Merge pull request #31 from mtandrita/feature/diabetes-ml-pipeline

Feature/diabetes ml pipeline

Author: shsarv4

Diabetes Prediction [END 2 END]/diabetes_pipeline/.gitignore

@@ -0,0 +1,4 @@
+__pycache__/
+*.pyc
+model/*.pkl
+.venv/

Diabetes Prediction [END 2 END]/diabetes_pipeline/README.md

@@ -0,0 +1,152 @@
+# Diabetes Prediction – Machine Learning Pipeline
+
+> ⚠️ This repository is a **forked project**.  
+> The work below represents my **independent contribution and extension** to the original codebase.
+
+This project implements a complete **end-to-end machine learning pipeline** for predicting diabetes using the Pima Indians Diabetes dataset.  
+The pipeline covers **data preprocessing, model training, evaluation, experimentation, and inference via CLI**.
+
+---
+
+## 📁 Project Structure
+diabetes_pipeline/
+│
+├── dataset/
+│ └── kaggle_diabetes.csv
+│
+├── model/
+│ ├── diabetes_model.pkl
+│ └── scaler.pkl
+│
+├── experiments/
+│ └── experiment_runner.py
+│
+├── data_preprocessing.py
+├── train.py
+├── predict.py
+├── evaluate.py
+└── README.md
+
+---
+
+## 🚀 My Contributions
+
+I independently designed and implemented the following components:
+
+### 1. Data Preprocessing Pipeline
+- Handled missing values in medical features:
+	- `Glucose`, `BloodPressure`, `SkinThickness`, `Insulin`, `BMI`
+- Replaced invalid zeros with `NaN`
+- Applied **mean / median imputation**
+- Standardized features using `StandardScaler`
+- Ensured consistent feature names across training and inference
+
+📄 `data_preprocessing.py`
+
+---
+
+### 2. Model Training
+- Implemented a reproducible training pipeline
+- Trained and persisted:
+	- Random Forest classifier
+	- Feature scaler
+- Stored trained artifacts for reuse and deployment
+
+📄 `train.py`
+
+---
+
+### 3. Model Evaluation
+- Added evaluation logic with:
+	- Accuracy
+	- Precision, Recall, F1-score
+- Verified generalization on the test set
+
+📄 `evaluate.py`
+
+---
+
+### 4. Experimentation Framework
+- Benchmarked multiple ML models:
+	- Logistic Regression
+	- Decision Tree
+	- Random Forest
+	- Support Vector Machine (SVM)
+- Automatically reports accuracy and F1-score
+
+📄 `experiments/experiment_runner.py`
+
+#### Sample Results
+
+| Model                 | Accuracy | F1 Score |
+|----------------------|----------|----------|
+| Logistic Regression  | 0.7875   | 0.6320   |
+| Decision Tree        | 0.9875   | 0.9805   |
+| Random Forest        | 0.9950   | 0.9921   |
+| SVM                  | 0.8450   | 0.7328   |
+
+✔️ **Random Forest performs best on this dataset**
+
+---
+
+### 5. Command-Line Prediction Interface
+- Built a CLI-based inference script
+- Ensures:
+	- Correct feature order
+	- Feature-name alignment with trained scaler
+- Predicts diabetes for a single patient input
+
+📄 `predict.py`
+
+Example:
+```bash
+python predict.py \
+	--pregnancies 2 \
+	--glucose 120 \
+	--bp 70 \
+	--skin 20 \
+	--insulin 80 \
+	--bmi 25 \
+	--dpf 0.5 \
+	--age 35
+
+
+
+---
+
+## 🛠️ Tech Stack
+
+- Python 3.10+
+- pandas
+- numpy
+- scikit-learn
+- joblib
+
+---
+
+## 🧩 Notes
+
+- Project is modular and deployment-ready
+- Structured to support FastAPI / Flask integration
+- Generated files cleaned using `.gitignore`
+- Suitable for internship-level ML engineering evaluation
+
+---
+
+## 👩‍💻 Author Contribution
+
+**Contributor:** Tandrita Mukherjee  
+
+**Contribution Scope:**
+- ML pipeline design
+- Data preprocessing
+- Model training & evaluation
+- Experimentation framework
+- CLI-based inference system
+
+---
+
+## 📌 Disclaimer
+
+This repository is a fork of an existing project.  
+All enhancements, restructuring, and ML pipeline components listed above were implemented independently as part of my learning and internship preparation.

Diabetes Prediction [END 2 END]/diabetes_pipeline/config.py

@@ -0,0 +1,7 @@
+from pathlib import Path
+
+BASE_DIR = Path(__file__).resolve().parent
+
+MODEL_DIR = BASE_DIR / "model"
+MODEL_PATH = MODEL_DIR / "diabetes_model.pkl"
+SCALER_PATH = MODEL_DIR / "scaler.pkl"

Diabetes Prediction [END 2 END]/diabetes_pipeline/data_preprocessing.py

@@ -0,0 +1,29 @@
+# diabetes_pipeline/data_preprocessing.py
+
+import pandas as pd
+import numpy as np
+from pathlib import Path
+from sklearn.model_selection import train_test_split
+
+def load_and_preprocess(test_size=0.2, random_state=0):
+    BASE_DIR = Path(__file__).resolve().parent
+    csv_path = BASE_DIR / "dataset" / "kaggle_diabetes.csv"
+    df = pd.read_csv(csv_path)
+
+    df = df.rename(columns={'DiabetesPedigreeFunction': 'DPF'})
+
+    cols_with_zero = ['Glucose','BloodPressure','SkinThickness','Insulin','BMI']
+    df[cols_with_zero] = df[cols_with_zero].replace(0, np.nan)
+
+    df['Glucose'] = df['Glucose'].fillna(df['Glucose'].mean())
+    df['BloodPressure'] = df['BloodPressure'].fillna(df['BloodPressure'].mean())
+    df['SkinThickness'] = df['SkinThickness'].fillna(df['SkinThickness'].median())
+    df['Insulin'] = df['Insulin'].fillna(df['Insulin'].median())
+    df['BMI'] = df['BMI'].fillna(df['BMI'].median())
+
+    X = df.drop(columns='Outcome')
+    y = df['Outcome']
+
+    return train_test_split(
+        X, y, test_size=test_size, random_state=random_state
+    )

… [END 2 END]/dataset/kaggle_diabetes.csv …tes_pipeline/dataset/kaggle_diabetes.csvDiabetes Prediction [END 2 END]/dataset/kaggle_diabetes.csv renamed to Diabetes Prediction [END 2 END]/diabetes_pipeline/dataset/kaggle_diabetes.csv Diabetes Prediction [END 2 END]/dataset/kaggle_diabetes.csv renamed to Diabetes Prediction [END 2 END]/diabetes_pipeline/dataset/kaggle_diabetes.csv

@@ -0,0 +1,20 @@
+import joblib
+from sklearn.metrics import accuracy_score, classification_report
+from data_preprocessing import load_and_preprocess
+from config import MODEL_PATH
+
+# Load data
+X_train, X_test, y_train, y_test, _ = load_and_preprocess()
+
+# Load trained model
+model = joblib.load(MODEL_PATH)
+
+# Predict
+y_pred = model.predict(X_test)
+
+# Metrics
+accuracy = accuracy_score(y_test, y_pred)
+report = classification_report(y_test, y_pred)
+
+print("Accuracy:", accuracy)
+print("\nClassification Report:\n", report)

Diabetes Prediction [END 2 END]/diabetes_pipeline/evaluate.py

@@ -0,0 +1,43 @@
+# diabetes_pipeline/experiments/experiment_runner.py
+
+import pandas as pd
+from sklearn.pipeline import Pipeline
+from sklearn.preprocessing import StandardScaler
+from sklearn.linear_model import LogisticRegression
+from sklearn.tree import DecisionTreeClassifier
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.svm import SVC
+from sklearn.metrics import accuracy_score, f1_score
+
+from diabetes_pipeline.data_preprocessing import load_and_preprocess
+
+X_train, X_test, y_train, y_test = load_and_preprocess()
+
+models = {
+	"LogisticRegression": LogisticRegression(max_iter=1000),
+	"DecisionTree": DecisionTreeClassifier(random_state=0),
+	"RandomForest": RandomForestClassifier(n_estimators=50, random_state=0),
+	"SVM": SVC()
+}
+
+results = []
+
+for name, model in models.items():
+	pipeline = Pipeline([
+		("scaler", StandardScaler()),
+		("model", model)
+	])
+
+	pipeline.fit(X_train, y_train)
+	preds = pipeline.predict(X_test)
+
+	results.append({
+		"Model": name,
+		"Accuracy": accuracy_score(y_test, preds),
+		"F1 Score": f1_score(y_test, preds)
+	})
+
+df = pd.DataFrame(results)
+print(df)
+
+df.to_csv("diabetes_pipeline/experiments/results.csv", index=False)

Diabetes Prediction [END 2 END]/diabetes_pipeline/experiments/__init__.py

@@ -0,0 +1,5 @@
+Model,Accuracy,F1 Score
+LogisticRegression,0.7875,0.6320346320346321
+DecisionTree,0.9875,0.980544747081712
+RandomForest,0.995,0.9921259842519685
+SVM,0.845,0.7327586206896551

Diabetes Prediction [END 2 END]/diabetes_pipeline/experiments/experiment_runner.py

@@ -0,0 +1,4 @@
+2025-12-28 11:48:56,518 - INFO - Training started
+2025-12-28 11:48:56,641 - INFO - Model and scaler saved successfully
+2025-12-28 11:49:14,730 - INFO - Training started
+2025-12-28 11:49:14,821 - INFO - Model and scaler saved successfully