Harriet-ngomo/Taxpayer-Risk-Classification

This project uses machine learning to classify taxpayers into risk levels (Low, High) based on financial and compliance data. It aims to help tax authorities improve audit targeting, reduce evasion, and support informed policy decisions. The workflow includes data cleaning, EDA, feature engineering, model building, and actionable recommendations # FairTax Analytics: Taxpayer Risk Classification Project - 🔗 Final Presentation ## Overview **FairTax Analytics** is a data driven company dedicated to transforming how tax authorities identify and prioritize high risk taxpayers. Tax evasion and non compliance undermine government revenue that is vital for essential public services like education, healthcare and infrastructure This project developed by FairTax Analytics leverages machine learning to automate taxpayer risk classification enabling smarter audit targeting early risk detection and more equitable enforcement ## Business and Data Understanding ### Business Problem Manual audit selection often relies on limited human judgment and scarce resources, leading to inefficiencies and unfair targeting. Our core business goal is to: > **Develop a predictive model to classify taxpayers by risk level (Low VS High)** to support targeted audits, enhance compliance monitoring and optimize revenue recovery ### Stakeholders | Stakeholder | Interest / Role | |-----------------------------|------------------------------------------------| | **Tax Authority / Revenue Service** | Maximize revenue, improve compliance | | **Audit Teams** | Prioritize audits efficiently | | **Policy Makers** | Inform regulation and policy design | | **Taxpayers** | Ensure fair compliance monitoring | | **Data Analysts / Scientists** | Build, maintain, and explain predictive models | ### Dataset Summary - **Data Source**: Tax Risk Identification Dataset (Kaggle) - **Observations**: 1000 - **Features**: 13 columns including financial and behavioral metrics - **Target Variable**: `Risk_Label` (High = 1, Low = 0).Medium risk entries were excluded for binary classification #### Key Features | Column Name | Description | Type | |--------------------- …