Portfolio - Prajwal Aher

Prajwal Aher

Data Aspirant

Resume

1

Machine Learning Project

About Me

I'm Prajwal Aher, a student at the College of Engineering Pune (COEP) with a strong interest in Data Science and a passion for turning raw data into meaningful insights.

Technical Skills
• Languages & Querying: Python, SQL
• Data Manipulation: Pandas, NumPy for cleaning, transforming, and analyzing structured datasets
• Machine Learning: Scikit-learn for building, training, and evaluating predictive models
• Data Visualization: Matplotlib, Seaborn for exploratory data analysis and communicating insights visually
• Cloud Platforms: AWS SageMaker for scalable model training and deployment workflows

Project Experience
• Built a Health Insurance Cost Predictor an end-to-end ML project covering data preprocessing, feature engineering, and model deployment
• Engineered a custom medical risk-scoring system to convert categorical health history into a normalized numerical feature

Key Skills

Python

SQL

Scikit-Learn

AWS Sagemaker

Streamlit

Pandas

NumPy

Matplotlib

Seaborn

My Projects

Healthcare Insurance Premium Predictor

Domain/Function: Healthcare

My Experience

Fresher

Awards & Certificate

Python: Beginner to Advanced For Data Professionals

Math and Statistics For AI, Data Science

Master Machine Learning for Data Science & AI: Beginner to Advanced

Let's Connect

Feel free to get in touch with me. I am always open to discussing new projects, creative ideas or opportunities to be part of your visions.

Download Resume

Resume

Mail Me

[email protected]

https://www.linkedin.com/in/prajwal-aher-4a2260320/

Call Me

+91 8446422281

GitHub

https://github.com/prajwalaher30-design

Machine Learning Project

Healthcare Insurance Premium Predictor

Domain/Function: Healthcare

Health Insurance Cost Predictor, An ML web app that estimates insurance premiums from demographic, medical, and lifestyle data using age-segmented XGBoost regression models, a custom risk-scoring feature, and real-time predictions deployed via Streamlit.

Data Cleaning: number_of_dependants found invalid negative values (-3, -1) from likely data entry errors. Fixed using .abs() to convert them to valid positive counts. Verified via df.describe(): values now range 0+, mean ~1.72 across 49,976 records.

Feature Importance (XGBoost) extracted via feature_importances_ and visualized. insurance_plan and age dominate predictions, followed by BMI, risk score, and smoking status. Most one-hot demographic features (region, gender) show minimal impact.

Project Details

Health Insurance Cost Predictor, An end-to-end ML web app that estimates annual health insurance premiums from demographic, medical, and lifestyle data.

• Built two age-segmented regression models (≤25 and >25), each with its own dedicated scaler, since cost patterns differ meaningfully across age groups
• Engineered a custom risk-scoring feature converting medical history (Diabetes, Heart Disease, Thyroid, and combinations) into a normalized 0–1 score
• Cleaned real-world data issues, including invalid negative values found in dependant counts, using validation checks
• Compared Logistic Regression, Random Forest, and XGBoost; selected the best model using evaluation metrics and feature importance analysis
• Found insurance_plan and age to be the strongest cost drivers, guiding feature prioritization
• Deployed as an interactive Streamlit app with a 12-field input form and real-time predictions

Tech Stack: Python, Pandas, NumPy, Scikit-learn, XGBoost, Matplotlib, Seaborn