Logo Lanfrica
  • Home
  • Atlas
  • Insights
  • Docs
  • Sign in

© 2026 Lanfrica. All rights reserved. All copyrights of the resources shown on the Lanfrica website belong to the original copyright holders, unless explicitly stated otherwise.

Alki45/insurance-risk-analytics-predictive-modeling

Domain:

socioeconomic

Record type:

project
Creator:
Alk
Host:
End-to-end analytics pipeline for car insurance risk segmentation, pricing optimization, and predictive modeling using South African insurance claims data. Built for AlphaCare Insurance Solutions (ACIS) during 10 Academy's AI Mastery Challenge. # Insurance Risk Analytics & Predictive Modeling This repository contains an end-to-end data science pipeline built during **10 Academy's AI Mastery Week 3 Challenge**. The goal is to help **AlphaCare Insurance Solutions (ACIS)** improve risk segmentation, optimize premium pricing, and identify low-risk, high-value customers using historical South African car insurance data. --- ## 📌 Project Overview This project addresses key business questions for ACIS by: - Analyzing and visualizing customer and claim data - Performing A/B testing to validate pricing strategies - Building regression and classification models to predict: - Expected annual premium (`regression`) - Risk level of a customer (`classification`) - Interpreting model outputs using SHAP - Managing reproducibility with DVC --- ## 📊 Key Components ### 1. 📁 Data - `train.csv`, `test.csv` – anonymized customer and claim-level data - Structured into: - Demographic information - Driving behavior - Claims and premium details ### 2. 🔍 Exploratory Data Analysis (EDA) - Univariate & bivariate analysis - Risk segmentation by age, claim history, gender, and region - Premium vs claim behavior visualization ### 3. 📈 Statistical Testing - A/B testing to evaluate if new pricing strategies outperform existing ones - Hypothesis testing on average premiums and risk levels ### 4. 🧠 Predictive Modeling - **Regression Model:** Predict customer's annual premium - **Classification Model:** Predict whether a customer is high- or low-risk - Model Evaluation: RMSE, Accuracy, ROC-AUC ### 5. 🔁 DVC (Data Version Control) - Tracks data, models, and outputs - Ensures reproducibility ### 6. 🧠 Model Interpretation - SHAP values to understand which features drive risk and premium predictions --- ## 🧪 Tech Stack | Tool / Library | Purpose | |------------------------|----------------------------------| | Python, pandas, NumPy | Data manipulation | | Matplotlib, seaborn | Vis …

Visit

github.com

Licenses

MIT