Medical Insurance Cost Prediction using Random Forest Regressor

Objective

To build a machine learning model that predicts medical insurance costs based on various features using a Random Forest Regressor. The model will be deployed as a web application using Flask.

Features and Methodology

1. Exploring the Dataset:
   • Understanding the dataset and its features.
   • Checking for missing values and data types.
2. Converting Categorical Values to Numerical:
   • Using techniques like one-hot encoding to convert categorical features into numerical values for the machine learning model.
3. Plotting Heatmap:
   • Visualizing the correlation between dependent and independent features using a heatmap.
4. Data Visualization:
   • Creating plots to visualize relationships between different features.
   • Plotting skewness and kurtosis to understand data distribution.
5. Data Preparation:
   • Splitting the dataset into training and testing sets.
   • Scaling the data if necessary.
6. Prediction Using Different Models:
   • Linear Regression
   • Support Vector Regressor (SVR)
   • Ridge Regressor
   • Random Forest Regressor
7. Hyperparameter Tuning:
   • Performing hyperparameter tuning to improve model performance.
8. Model Comparison:
   • Plotting graphs to compare the performance of different models.
9. Model Deployment:
   • Preparing the model for deployment.
   • Deploying the model using Flask.

Results

The Random Forest Regressor achieved an accuracy of 86% for predicting medical insurance costs.

Dataset

The dataset used for this project can be downloaded from Kaggle. Click to Download.

Installation Steps

1. Install Python 3.7.0.
2. Install Dependencies

   python -m pip install --user -r requirements.txt

3. Run the Application

   python app.py