ML Workbench

An interactive Machine Learning Dashboard built with Streamlit that enables users to perform Classification, Regression, and Clustering workflows through a no-code graphical interface. The application supports dataset exploration, preprocessing, model training, performance evaluation, visualization, and Excel report generation.

Streamlit App : Try the interactive dashboard

---

Project Overview

Feature	Description
Dataset Exploration	Load built-in datasets or upload custom datasets
Data Preprocessing	Missing value handling, encoding, scaling, and cleaning
Classification	Train and evaluate multiple classification algorithms
Regression	Train and evaluate regression models
Clustering	Perform unsupervised clustering analysis
Visualization Studio	Interactive plots and performance visualizations
Excel Export	Export model results and evaluation reports
Interactive UI	Built using Streamlit for an intuitive workflow

Project Architecture

Module	Responsibility
app.py	Streamlit user interface and workflow orchestration
datasets.py	Dataset loading and management
preprocessing.py	Data cleaning, encoding, scaling, and preprocessing
models.py	Machine learning model registry and training
metrics.py	Evaluation metric computation
visualization.py	Plot generation and visual analytics
export_excel.py	Excel report creation and export

Project Structure

ML-Workbench/
│
├── app.py
├── datasets.py
├── models.py
├── metrics.py
├── visualization.py
├── preprocessing.py
├── export_excel.py
│
├── README.md
├── requirements.txt

Technology Stack

Category	Technologies
Language	Python
Frontend	Streamlit
Machine Learning	Scikit-Learn, XGBoost
Data Processing	Pandas, NumPy
Visualization	Plotly, Matplotlib, Seaborn
Export Engine	OpenPyXL
Dimensionality Reduction	PCA, UMAP
Explainability Ready	SHAP

---

Supported Machine Learning Tasks

Task Type	Available
Classification	Logistic Regression, Decision Tree, Random Forest, KNN, Naive Bayes, SVM, LDA, QDA, AdaBoost, Gradient Boosting, XGBoost, Bagging, Extra Trees
Regression	Linear Regression, Ridge, Lasso, ElasticNet, Decision Tree Regressor, Random Forest Regressor, KNN Regressor, SVR, AdaBoost Regressor, Gradient Boosting Regressor, XGBoost Regressor, Bagging Regressor
Clustering	KMeans, Agglomerative Clustering, DBSCAN, MeanShift, Birch, Gaussian Mixture Models

---

Built-in Datasets

Dataset	Category
Iris	Classification
Wine	Classification
Breast Cancer	Classification
Titanic	Classification
Penguins	Classification
Diabetes	Regression
California Housing	Regression
MPG	Regression
Tips	Regression
Mall Customers	Clustering

---

Metrics

Classification Metrics
Accuracy
Precision
Recall
F1 Score
Specificity
Sensitivity
True Positive Rate (TPR)
False Positive Rate (FPR)
ROC-AUC Score
Confusion Matrix

---

Regression Metrics

Metric
Mean Absolute Error (MAE)
Mean Squared Error (MSE)
Root Mean Squared Error (RMSE)
R² Score
Adjusted R²
Mean Absolute Percentage Error (MAPE)

---

Clustering Metrics

Metric
Silhouette Score
Davies-Bouldin Index
Calinski-Harabasz Index
Inertia

---

Visualization Features

Classification

• Confusion Matrix
• ROC Curve
• Precision Recall Curve
• Feature Importance

Regression

• Actual vs Predicted
• Residual Plot
• Error Distribution

Clustering

• Cluster Visualization
• Elbow Method
• Silhouette Analysis

General

• Correlation Heatmap
• PCA Projection

---

Installation

Clone Repository

git clone https://github.com/karmakar-rahul/ML-Workbench.git
cd ML-Workbench

Install Dependencies

pip install -r requirements.txt

Run Application

streamlit run app.py

---

Future Improvements

• Hyperparameter Optimization
• SHAP Explainability
• Model Comparison Dashboard
• AutoML Workflow
• Feature Selection Tools
• Time Series Analysis

---

Author

Rahul Karmakar

• PGCP - BDA , C-DAC Chenai
• M.Sc. Physics (Astrophysics), Assam University

---

License

This project is released under the MIT License.