Project: CareerFoundry Data Immersion Program – Achievement 5
Objective: Provide analytical support to a global bank's anti-money-laundering compliance department, assessing client and transaction risk while handling real-world data ethics challenges.

• What factors contribute most to a client's likelihood of closing their account?
• How can we identify and mitigate bias in client risk models?
• How do ethical considerations affect the collection, use, and sharing of client data?
• How can time-series forecasting support compliance reporting and decision-making?

• Client dataset including demographics, account activity, and product usage
• Analyzed using Microsoft Excel
• Decision tree model built to predict account churn risk

• Microsoft Excel for data cleaning, descriptive statistics, and time-series analysis
• Decision tree modeling for churn prediction
• Linear regression for predictive analysis
• Time-series analysis & forecasting (moving averages)
• Data ethics frameworks: bias identification, security & privacy considerations
• GitHub for portfolio hosting and version control

1. Big Data Exploration: Identified characteristics of structured vs. unstructured data and limitations of big data approaches
2. Data Ethics – Bias: Identified sources of bias in the client dataset and proposed mitigation strategies
3. Data Ethics – Security & Privacy: Analyzed ethical dilemmas and relevant data protection considerations
4. Data Mining: Cleaned client data, computed descriptive statistics, and built a decision tree to model churn outcomes
5. Predictive Analysis: Applied linear regression concepts to client risk scenarios
6. Time-Series Analysis: Created moving averages in Excel and explored forecasting models
7. Reporting & Portfolio: Documented findings and hosted work on GitHub as part of a data analytics portfolio

• 📊 Client Data Set Analysis (Excel) — available in this repository
• 🌳 Decision Tree – Churn Model (PDF) — available in this repository

• Key churn risk factors identified include account age, activity status, number of products held, and client demographics
• Decision tree model highlights the compounding risk when multiple churn indicators are present
• Bias analysis revealed the importance of reviewing demographic variables in compliance models
• Time-series patterns in client activity provide a basis for proactive risk forecasting

• Apply churn model to flag at-risk clients for proactive retention outreach
• Review model variables regularly to control for demographic bias
• Strengthen data governance practices around client PII in compliance workflows
• Use time-series forecasting to anticipate volume changes in compliance workloads

Author: Gabriela Cascione