ML Binary & Multi-Class Classification Project

A comprehensive Machine Learning classification pipeline covering 6 datasets with 6 ML models each, totaling 36 trained models.

Datasets

#	Dataset	Type	Samples	Features
1	Breast Cancer	Binary	569	30
2	Pima Diabetes	Binary	768	8
3	Titanic	Binary	891	11
4	Iris	Multi-Class	150	4
5	Wine Quality	Multi-Class	1,599	11
6	Dry Bean	Multi-Class	13,611	16

Models Used

• Logistic Regression
• Decision Tree
• Random Forest
• Support Vector Machine (SVM)
• K-Nearest Neighbors (KNN)
• XGBoost

Complete ML Pipeline

Step	Description
EDA	Distributions, correlations, pairplots, boxplots
Outlier Handling	IQR detection + Winsorization (capping)
Missing Values	Median imputation for biologically impossible zeros
Feature Engineering	Title extraction, FamilySize, IsAlone (Titanic)
Feature Selection	ANOVA F-test + Mutual Information scoring
Class Imbalance	SMOTE (Synthetic Minority Over-sampling)
Feature Scaling	StandardScaler (mean=0, std=1)
Cross-Validation	5-Fold Stratified CV for model stability
Evaluation	Accuracy, Precision, Recall, F1, ROC-AUC
Classification Reports	Per-class precision, recall, F1
Feature Importance	Tree-based model feature ranking
Learning Curves	Overfitting/underfitting visualization
Hyperparameter Tuning	GridSearchCV with 5-fold CV
Model Saving	Best models exported with joblib

How to Run

1. Open Google Colab
2. Upload ML_Classification_Complete.ipynb
3. Upload all dataset files when prompted
4. Run all cells (~10-15 min)

Tech Stack

• Python 3.x
• scikit-learn
• XGBoost
• imbalanced-learn (SMOTE)
• pandas, numpy
• matplotlib, seaborn

Project Structure

├── ML_Classification_Complete.ipynb   # Main notebook (Google Colab)
├── ML_Classification_Complete.py      # Python script version
├── data.csv                           # Breast Cancer dataset
├── diabetes.csv                       # Pima Diabetes dataset
├── titanic/                           # Titanic dataset
│   ├── train.csv
│   ├── test.csv
│   └── gender_submission.csv
├── Iris/
│   └── Iris.csv                       # Iris dataset
├── winequality-red.csv                # Wine Quality dataset
├── Dry_Bean_Dataset/
│   ├── Dry_Bean_Dataset.xlsx          # Dry Bean dataset
│   └── Dry_Bean_Dataset.arff
└── README.md