Shegaw-21hub/ACIS_Insurance_Analytics

This repository supports ACIS Insurance Analytics at AlphaCare Insurance Solutions, applying machine learning and statistical modeling to enhance car insurance planning in South Africa. It includes risk analysis, A/B testing, predictive modeling, and optimized premium pricing to improve marketing strategies and customer targeting. # AlphaCare Insurance Solutions Risk Analytics ## Project description This project analyzes historical car insurance claim data in South Africa to optimize marketing and pricing strategies. The analysis includes EDA, hypothesis testing, and predictive modeling. ## Tasks: - EDA & Stats - DVC Pipeline - A/B Hypothesis Testing - Machine Learning Modeling ### Structure - /data: Raw & processed data - /notebooks: Jupyter notebooks - /scripts: Reusable Python code - /outputs: Plots & results ### Requirements - pandas - seaborn - matplotlib - scikit-learn - statsmodels - dvc # 📊 Task 1: Git Setup & Exploratory Data Analysis (EDA) --- ## 📌 Objective This task focuses on establishing a solid foundation for the project by: - Setting up **version control** using Git and GitHub with CI/CD via GitHub Actions. - Performing a comprehensive **Exploratory Data Analysis (EDA)** to understand the insurance dataset. - Applying statistical methods and visualizations to uncover insights on risk and profitability. --- ## 🛠️ What Was Done ### 1. Git and GitHub Setup - Created a Git repository dedicated to the project. - Established a `task-1` branch to isolate analysis work. - Implemented continuous integration using **GitHub Actions** to automate code checks and workflows. - Committed work regularly with descriptive messages to maintain clear version history. ### 2. Exploratory Data Analysis (EDA) & Statistics - **Data Understanding:** Reviewed dataset structure, verified data types, and assessed data quality including missing values. - **Descriptive Statistics:** Calculated variability and central tendencies for key numeric variables such as `TotalPremium` and `TotalClaims`. - **Univariate Analysis:** Visualized distributions using histograms (numerical data) and bar charts (categorical data). - **Bivariate & Multivariate Analysis:** Explored relationships between variables, including monthly changes in premiums and claims across ZipCodes, using scatter plots and correlat …