https://github.com/hariprasath-v/av-dataverse-hack---insurance-claim-prediction

Create a machine learning model to predict if the policyholder will file a claim in the next 6 months or not based on the set of car and policy features.
https://github.com/hariprasath-v/av-dataverse-hack---insurance-claim-prediction

analyticsvidhya classification exploratory-data-analysis f1-score matplotlib numpy pandas python randomforest-classification scikit-learn seaborn shap

Last synced: 12 days ago
JSON representation

Create a machine learning model to predict if the policyholder will file a claim in the next 6 months or not based on the set of car and policy features.

• Host: GitHub
• URL: https://github.com/hariprasath-v/av-dataverse-hack---insurance-claim-prediction
• Owner: hariprasath-v
• License: apache-2.0
• Created: 2022-11-16T14:14:18.000Z (about 2 years ago)
• Default Branch: main
• Last Pushed: 2022-11-16T14:33:21.000Z (about 2 years ago)
• Last Synced: 2024-11-13T15:54:32.191Z (2 months ago)
• Topics: analyticsvidhya, classification, exploratory-data-analysis, f1-score, matplotlib, numpy, pandas, python, randomforest-classification, scikit-learn, seaborn, shap
• Language: Jupyter Notebook
• Homepage:
• Size: 7.36 MB
• Stars: 0
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# AV-Dataverse-Hack---Insurance-Claim-Prediction

### Competition hosted on Analyticsvidhya

# About

### Create a machine learning model to predict if the policyholder will file a claim in the next 6 months or not based on the set of car and policy features.

### Final Competition score is 0.1749408983

### Leaderboard Rank is 38

### Evaluation Metric is F1-score.

### File information

 

 * av-dataverse-hack-insurance-claim-prediction-eda.ipynb [![Open in Kaggle](https://img.shields.io/static/v1?label=&message=Open%20in%20Kaggle&labelColor=grey&color=blue&logo=kaggle)](https://www.kaggle.com/hari141v/av-dataverse-hack-insurance-claim-prediction-eda)

    #### Basic Exploratory Data Analysis

    #### Packages Used,

        * seaborn

        * Pandas

        * Numpy

        * Matplotlib

* av-dataverse-hack-insurance-claim-prediction-model.ipynb [![Open in Kaggle](https://img.shields.io/static/v1?label=&message=Open%20in%20Kaggle&labelColor=grey&color=blue&logo=kaggle)](https://www.kaggle.com/hari141v/av-dataverse-hack-insurance-claim-prediction-model)

    #### Data Pre-processing and model. 

    #### Packages Used,

        * Sklearn

        * Pandas

        * Numpy

        * Matplotlib

        * shap

     #### Created Random forest classifier model and evaluated with f1-score. 

     #### [For more detailed information about the model.](https://github.com/hariprasath-v/AV-Dataverse-Hack---Insurance-Claim-Prediction/blob/main/Approach_AV%20Dataverse%20Hack%20-%20Insurance%20Claim%20Prediction.pdf)

     

### Random Forest Model Feature Importances

![Alt text](https://github.com/hariprasath-v/AV-Dataverse-Hack---Insurance-Claim-Prediction/blob/main/Model%20Visualization/Randomforest%20model%20-%20top%2010%20feature%20importances.png)

### Random Forest Model - SHAP Feature Importances

![Alt text](https://github.com/hariprasath-v/AV-Dataverse-Hack---Insurance-Claim-Prediction/blob/main/Model%20Visualization/Randomforest%20model%20-%20SHAP%20top%2010%20feature%20importances.png)

### SHAP Top feature influences the class 0

![Alt text](https://github.com/hariprasath-v/AV-Dataverse-Hack---Insurance-Claim-Prediction/blob/main/Model%20Visualization/Randomforest%20model%20-%20SHAP%20Top%20feature%20influences%20the%20class%200.png)

### SHAP Top feature influences the class 1

![Alt text](https://github.com/hariprasath-v/AV-Dataverse-Hack---Insurance-Claim-Prediction/blob/main/Model%20Visualization/Randomforest%20model%20-%20SHAP%20Top%20feature%20influences%20the%20class%201.png)

### Threshold Tuning Results - Optimal threshold: 0.0745, F1-score: 0.16929

![Alt text](https://github.com/hariprasath-v/AV-Dataverse-Hack---Insurance-Claim-Prediction/blob/main/Model%20Visualization/Probability%20Threshold%20Tuning%20-%20Optimal%20threshold%200.0745%2C%20F1-score%20%200.16929.png)