🧠 End-to-End Machine Learning Project Report
📌 Project Overview
This project involved building and evaluating machine learning models to predict
The full ML lifecycle was covered, including data preprocessing, modeling, evaluation, and explainability.
🛠️ Workflow Summary
✅ Data Preprocessing
- Handled missing values using
IterativeImputer. - Removed outliers with a custom
OutlierClipper. - Scaled numerical features using
StandardScaler. - Selected top features via
SelectKBestwithf_regression.
✅ Data Splitting
- Split into train (68%), validation (17%), and holdout (15%) sets using
train_test_split. - Ensured no data leakage by fitting preprocessors only on training data within the pipeline.
🤖 Model Training & Evaluation
The following models were trained using Repeated K-Fold Cross-Validation:
- Linear Models:
LinearRegression,Ridge,Lasso,ElasticNet - Tree-Based:
RandomForest,ExtraTrees,GradientBoosting,XGBoost - Others:
KNN,MLPRegressor
📊 Metrics Evaluated
- R² Score
- MAE
- RMSE
All results were compiled and visualized to compare model performances.
