-
Insurance Predictor
Inspiration
The idea came from analyzing how insurance companies estimate health risk and pricing. I wanted to replicate that logic using machine learning — building a transparent system that not only predicts costs but also helps users understand why certain features affect pricing.
What it does
I developed a Streamlit-based web application that takes user input or CSV data, performs automatic preprocessing (encoding, scaling, PCA), and applies machine learning models to predict insurance charges. The app also includes an EDA dashboard with correlation heatmaps, data summaries, and trend plots.
How I built it
Performing end-to-end ML workflow: data cleaning → feature engineering → model training → deployment. Visualizing relationships between variables using EDA. Understanding how different ML algorithms behave in regression tasks. Deploying an ML model interactively using Streamlit for user-friendly UI. Managing errors and optimizing code structure for readability.
Challenges I ran into
Upload a dataset or enter individual details. The app preprocesses the data (handles nulls, encodes categorical variables, scales features). Trains and evaluates regression models (e.g., Linear Regression, Decision Tree). Predicts insurance costs instantly and visualizes feature relationships.
Accomplishments that I'm proud of
A fully functional, self-developed ML application that simulates real-world insurance pricing logic and enhances interpretability through analytics and visual feedback.
