Skip to content
BranchesTags

Repository files navigation

Insurance Risk Analytics

Exploratory Data Analysis (EDA) and Machine Learning for AlphaCare Insurance Solutions
Data: MachineLearningRating_v3.csv (converted from pipe-separated .txt, Feb 2014–Aug 2015)

Table of Contents

Project Overview

This project analyzes insurance policy and claims data to:

  • Calculate and visualize loss ratios by province, vehicle type, and gender
  • Explore distributions and outliers in premiums, claims, and vehicle values
  • Identify temporal trends in claim frequency and severity
  • Highlight vehicle makes/models associated with high or low claim amounts
  • Build and evaluate predictive models for claim severity, claim probability, and premium optimization
  • Implement a dynamic, risk-based pricing system using machine learning
  • Interpret model results and provide actionable business insights for insurance pricing

Folder Structure

insurance-risk-analytics/
│
├── data/                   # Raw and processed data files
│   └── MachineLearningRating_v3.csv
├── notebooks/              # Jupyter notebooks for EDA and analysis
│   └── eda_analysis.ipynb
├── src/                    # Source code (data loading, preprocessing, utils)
│   ├── data_loader.py
│   ├── data_preprocessor.py
│   └── utils.py
├── plots/                  # Generated plots and visualizations
├── README.md
└── requirements.txt

Setup & Installation

  1. Clone the repository:

    git clone https://github.com/yourusername/insurance-risk-analytics.git
cd insurance-risk-analytics

  2. Create and activate a virtual environment (optional but recommended):

    python -m venv venv
venv\Scripts\activate  # On Windows

  3. Install dependencies:

    pip install -r requirements.txt

Usage

  • Run EDA notebook:
    1. Open notebooks/eda_analysis.ipynb in Jupyter or VS Code.
    2. Run all cells to generate summary statistics and plots.

=======

Exploratory Data Analysis (EDA) and Machine Learning for AlphaCare Insurance Solutions
Data: MachineLearningRating_v3.csv (converted from pipe-separated .txt, Feb 2014–Aug 2015)

Table of Contents

Project Overview

This project analyzes insurance policy and claims data to:

  • Calculate and visualize loss ratios by province, vehicle type, and gender
  • Explore distributions and outliers in premiums, claims, and vehicle values
  • Identify temporal trends in claim frequency and severity
  • Highlight vehicle makes/models associated with high or low claim amounts

Folder Structure

insurance-risk-analytics/
│
├── data/                   # Raw and processed data files
│   └── MachineLearningRating_v3.csv
├── notebooks/              # Jupyter notebooks for EDA and analysis
│   └── eda_analysis.ipynb
├── src/                    # Source code (data loading, preprocessing, utils)
│   ├── data_loader.py
│   ├── data_preprocessor.py
│   └── utils.py
├── plots/                  # Generated plots and visualizations
├── README.md
└── requirements.txt

  • Convert raw TXT to CSV (if needed):

    python src/data_loader.py

  • Preprocess data for modeling:

    python src/data_preprocessor.py

  • Run Predictive Modeling notebook (Task 4):

    1. Open notebooks/predictive_modeling.ipynb.
    2. Run all cells to build, evaluate, and interpret predictive models for risk-based pricing.
    3. Review outputs for model performance, feature importance, and business recommendations.

Key Features

  • Data Cleaning: Handles missing values, mixed types, and converts columns to appropriate types.
  • Descriptive Statistics: Summarizes key numerical and categorical features.
  • Visualization: Generates histograms, bar charts, scatter plots, box plots, and heatmaps.
  • Loss Ratio Analysis: Calculates and visualizes loss ratios by region and other factors.
  • Temporal Analysis: Examines trends in claims over time.
  • Vehicle Analysis: Identifies makes/models with highest and lowest claim amounts.
  • Predictive Modeling: Machine learning models for claim severity (regression), claim probability (classification), and premium optimization (regression).
  • Risk-Based Pricing Engine: Calculates premiums based on predicted claim probability and severity, with expense loading and profit margin.
  • Model Evaluation: Comprehensive metrics (RMSE, R², accuracy, precision, recall, F1-score, confusion matrix) and cross-validation.
  • Interpretability: SHAP analysis for feature importance and business interpretation.
  • Business Insights: Actionable recommendations for pricing, risk segmentation, and model deployment.

Data Notes

  • CapitalOutstanding had comma-separated numbers, converted to floats.
  • MaritalStatus, Gender, CrossBorder had mixed types, set as categories.
  • Negative TotalPremium and TotalClaims may indicate refunds or errors.
  • CustomValueEstimate has 78% missing values.
  • Some columns may have mixed types; see code for dtype handling.

Results & Outputs

  • Plots: Saved in the plots/ directory (e.g., total_claims_hist.png, loss_ratio_province.png).
  • Summary statistics: Printed in the EDA notebook output.
  • Cleaned data: Saved as CSV after preprocessing.
  • Predictive modeling results:
    • Model performance metrics (RMSE, R², accuracy, F1, etc.)
    • SHAP feature importance and business interpretation
    • Risk-based premium calculations and comparisons
    • Business insights and recommendations for insurance pricing

Contributing

Pull requests and suggestions are welcome!
Please open an issue to discuss your ideas or report bugs.

License

=======

Setup & Installation

  1. Clone the repository:

    git clone https://github.com/yourusername/insurance-risk-analytics.git
cd insurance-risk-analytics

  2. Create and activate a virtual environment (optional but recommended):

    python -m venv venv
venv\Scripts\activate  # On Windows

  3. Install dependencies:

    pip install -r requirements.txt

Usage

  • Run EDA notebook:

    1. Open notebooks/eda_analysis.ipynb in Jupyter or VS Code.
    2. Run all cells to generate summary statistics and plots.

  • Convert raw TXT to CSV (if needed):

    python src/data_loader.py

  • Preprocess data for modeling:

    python src/data_preprocessor.py

Key Features

  • Data Cleaning: Handles missing values, mixed types, and converts columns to appropriate types.
  • Descriptive Statistics: Summarizes key numerical and categorical features.
  • Visualization: Generates histograms, bar charts, scatter plots, box plots, and heatmaps.
  • Loss Ratio Analysis: Calculates and visualizes loss ratios by region and other factors.
  • Temporal Analysis: Examines trends in claims over time.
  • Vehicle Analysis: Identifies makes/models with highest and lowest claim amounts.

Data Notes

  • CapitalOutstanding had comma-separated numbers, converted to floats.
  • MaritalStatus, Gender, CrossBorder had mixed types, set as categories.
  • Negative TotalPremium and TotalClaims may indicate refunds or errors.
  • CustomValueEstimate has 78% missing values.
  • Some columns may have mixed types; see code for dtype handling.

Results & Outputs

  • Plots: Saved in the plots/ directory (e.g., total_claims_hist.png, loss_ratio_province.png).
  • Summary statistics: Printed in the EDA notebook output.
  • Cleaned data: Saved as CSV after preprocessing.

Contributing

Pull requests and suggestions are welcome!
Please open an issue to discuss your ideas or report bugs.

License

MIT License

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

1 star

Watchers

0 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors

Languages