Machine Learning Lab

Supervised Classification Agriculture Random Forest

SmartHarvest – Crop Recommendation Engine

End‑to‑end ML system recommending the most suitable crop using soil nutrients (NPK), climate, and soil properties.

Stack: Python · Scikit‑learn · Pandas · Streamlit
Highlights: ~95% accuracy with feature‑engineered agronomic variables.

Live App View Code

Supervised Classification Healthcare KNN

Mr. Cardio Disease Astrologer

KNN‑based heart disease prediction using indicators like cholesterol, blood pressure and heart rate.

Stack: Python · Scikit‑learn · Streamlit
Highlights: Interactive UI for risk estimation with explainable inputs.

Live App View Code

Supervised Classification Finance Naïve Bayes

Lending Logic – Loan Approval System

Gaussian Naïve Bayes model predicting loan approval probability using engineered financial features.

Stack: Python · Scikit‑learn · Streamlit
Highlights: ~92% accuracy with focus on reducing false positives.

Live App View Code

Supervised Classification Healthcare SVM

CardioPredict – SVM Heart Risk Model

Support Vector Machine model predicting heart disease risk from clinical features with a Streamlit front-end.

Stack: Python · Scikit‑learn · Streamlit
Highlights: Margin‑based classifier tuned for medical risk prediction.

Live App View Code

Supervised Classification Marketing KNN

Mobile Market Segmenter

KNN model classifying mobile users into market segments using usage behaviour and demographic patterns.

Stack: Python · Scikit‑learn · Streamlit
Highlights: Simple, interpretable clustering-style segmentation via supervised labels.

Live App View Code

Supervised Classification Behavior Logistic Regression

Personality Type Predictor

Logistic Regression model mapping behavioural traits to personality categories with an interactive UI.

Stack: Python · Scikit‑learn · Pandas · Streamlit
Highlights: Clean feature pipeline for survey‑style inputs.

Live App View Code

Supervised Regression Mobility Linear Regression

Ride Price Predictor

Linear Regression estimator for ride fares using distance, duration, time of day and ride type.

Stack: Python · Scikit‑learn · Pandas · Streamlit
Highlights: Baseline pricing model for ride‑hailing style apps.

Live App View Code

Supervised Regression HR / Salary KNN Regression

Experience to Earnings

KNN regression model predicting salary from years of experience as an intuitive career analytics tool.

Stack: Python · Scikit‑learn · Streamlit
Highlights: Non‑linear salary curve captured via neighbor‑based regression.

Live App View Code

Supervised Regression Education KNN

Student Performance Analyzer

End-to-end ML regression model predicting student exam scores using KNN with hyperparameter tuning and feature engineering.

Stack: Python · Scikit-learn · Streamlit · Pandas
Highlights: MAE: 6.89 | RMSE%: ~17.44% | Feature importance analysis with model interpretability.

Live App View Code

Supervised Classification Behavioral Logistic Regression

PERSONA PULSE – MBTI Personality Classification

End-to-end ML pipeline for MBTI personality type classification using questionnaire data with leakage-safe scikit-learn Pipeline.

Stack: Python · Scikit-learn · Pandas · Streamlit
Highlights: Strong multiclass Logistic Regression (~0.92 F1) | Full model interpretability | Real-time predictions MVP app

Live App View Code

Supervised Regression & Classification E-commerce Ridge & Logistic

🚚 Delivery Oracle – E-commerce Delivery Intelligence

End-to-end delivery intelligence system using Olist dataset with Ridge regression for ETA prediction and Logistic Regression for late-delivery risk classification.

Stack: Python · Scikit-learn · Pandas · Streamlit · Joblib
Highlights: Multi-table feature engineering (distance, time, parcel metrics) | Streamlit checkout-style UI | Real-time delivery risk alerts and ETA predictions

Live App View Code

Supervised Regression E-commerce Random Forest

Used Car Price Predictor – CarDekho

Regression-based ML app using Random Forest Regressor to predict fair selling prices of used cars from 45K+ CarDekho listings with strong performance (R² ≈ 0.92).

Stack: Python · Scikit-learn · Streamlit · Pandas · Plotly

Features: Real-time price prediction from car specs | Feature importance and model insights dashboard | Handles categorical & numerical features with full preprocessing pipeline | MAE, RMSE, R² evaluation

Live App View Code

Supervised Classification Healthcare Decision Tree

Cardiovascular Health Risk Assessment

Interpretable ML system using a Decision Tree Classifier on 70K+ patient records to predict cardiovascular disease risk from vitals, lab values, and lifestyle factors.

Stack: Python · Scikit-learn · Streamlit · Pandas · Plotly · Joblib

Features: BMI and age-group based feature engineering | Real-time risk prediction with probability scores | Feature importance and clinical visualizations | Streamlit dashboard optimized for healthcare analytics

Live App View Code

Supervised Classification Healthcare Decision Tree

Medicine Recommendation System

AI-powered diagnostic tool using Decision Tree Classifier to predict diseases from symptoms and provide personalized recommendations.

Stack: Python · Scikit-learn · Streamlit · Pandas · Joblib

Features: Symptom-based disease prediction | Personalized medication, diet, precautions, and workout recommendations | Interactive Streamlit interface

Live App View Code

Supervised Classification Sports Analytics Random Forest

ATP Tennis Match Outcome Classifier

ML-powered sports analytics tool using Random Forest to predict ATP tennis match winners from pre-match rankings across 59K+ matches with 99.5% accuracy.

Stack: Python · Scikit-learn · Streamlit · Pandas · Plotly

Features: Ranking-based, leakage-free pre-match predictions | Win probability visualization | Analysis of 20 years of ATP history (2000–2019) | Interactive Streamlit interface

Live App View Code

Supervised Regression Finance Decision Tree

OHLCV Next-Day Close Predictor

Time-series ML mini product that predicts next-day stock closing prices using OHLCV features, Decision Tree Regression, and a naive “tomorrow ≈ today” baseline for comparison.

Stack: Python · Scikit-learn · Streamlit · Pandas · Plotly · yfinance · Joblib

Features: Real-time OHLCV fetch from Yahoo Finance | Lag, moving average, and volatility feature engineering | GridSearchCV with TimeSeriesSplit | MAE/RMSE vs naive baseline | Wall Street-themed Streamlit UI

Live App View Code

Supervised Regression Energy Random Forest

PJM Energy Demand Forecaster

End-to-end ML system predicting hourly electricity demand using 10+ years of PJM load data with Random Forest Regression and advanced time-series feature engineering.

Stack: Python · Scikit-learn · Streamlit · Pandas · Plotly · NumPy · Joblib

Highlights: 10+ years of multi-region PJM data (AEP, COMED, DAYTON, DEOK, DOM) | Time-based features (hour, day, month, season, lags) | MAE ≈ 500 MW, RMSE ≈ 700 MW, R² ≈ 0.95 | Interactive Streamlit dashboard for real-time forecasting and historical analysis

Live App View Code

Supervised Classification Healthcare SVM

VociPark – Parkinson’s Disease Detection

End-to-end Parkinson’s detection from voice data using an SVM classifier optimized via GridSearchCV and StandardScaler pipelines.

Stack: Python · Scikit-learn · Streamlit · Pandas · Joblib

Highlights: ~82% Test Accuracy | Balanced Accuracy ~0.80 | Stratified Train-Test Split | Tuned C, Gamma & Kernel

Live App View Code