Predicting Customer Interest in Health Insurance Cross-Sell

Problem Statement

In the competitive insurance industry, maximizing revenue through cross-selling additional products to existing customers is a key strategy. This project focuses on predicting customer interest in purchasing a vehicle insurance policy, specifically targeting those who already have a health insurance policy. Accurately identifying potential customers for this cross-sell opportunity is essential for optimizing marketing efforts, reducing customer acquisition costs, and enhancing overall customer satisfaction.

Objective

The primary objective of this project is to develop a machine learning model that predicts the likelihood of a customer being interested in purchasing a vehicle insurance policy. The model will leverage various customer attributes, including demographics, vehicle details, insurance history, and interaction channels, to make these predictions. The key target variable is Response, which indicates whether or not a customer is interested in the cross-sell offer.

Key Challenges

• Data Imbalance: The dataset may have an unequal distribution of interested versus uninterested customers, potentially leading to biased model predictions.
• Feature Engineering: Creating and identifying relevant features that can enhance model accuracy.
• Scalability: Ensuring the model's efficiency in handling large datasets.
• Model Evaluation: Choosing appropriate metrics, such as precision, to accurately measure the model's ability to predict true positives while minimizing false positives.

Deliverables

• A machine learning model capable of predicting the likelihood of a customer responding positively to a vehicle insurance cross-sell offer.
• A comprehensive report detailing:
  • Data preprocessing steps
  • Feature engineering techniques
  • Model selection and optimization
  • Performance evaluation using appropriate metrics
  • Key business insights derived from the model's predictions

Business Impact

Accurately predicting customer interest allows the insurance company to:

• Better allocate marketing resources to target the most promising customers.
• Increase the conversion rate of cross-sell campaigns.
• Enhance operational efficiency by focusing efforts on customers most likely to respond positively.
• Improve overall customer experience by offering relevant products to those who would benefit the most.

Project Structure

The repository is organized as follows:

• Machine_Learninig_Capstone_Project.ipynb: The main Jupyter notebook containing the code, data processing steps, model training, evaluation, and deployment instructions.
• data/: Directory containing the datasets used for training and testing the model.
• models/: Pre-trained models and saved model files.
• requirements.txt: List of Python libraries required to run the notebook and replicate the results.
• README.md: This file, providing an overview and instructions for the project.

Usage

To run the notebook:

1. Ensure your environment is set up as described in the Installation section.
2. Launch Jupyter Notebook:

   jupyter notebook

3. Open the Machine_Learninig_Capstone_Project.ipynb file and run the cells sequentially.

Model Performance

The final model achieved the following performance metrics:

• Accuracy: 94.89%
• Recall: 99.83%
• Precision: 90.87%
• F1-Score: 95.14%

These metrics demonstrate the model's effectiveness in predicting customer interest and its readiness for deployment in a real-world setting.

Conclusion

This project successfully developed a machine learning model that predicts customer interest in a vehicle insurance cross-sell offer. By leveraging this model, the insurance company can enhance marketing efficiency, improve customer targeting, and ultimately increase the conversion rate of their cross-sell campaigns.

License

This project is licensed under the MIT License.

Acknowledgments

• Special thanks to AlmaBetter for providing resources and support.