Individual Project

Key steps in the case study.<ol><li>Data collection and Data preprocessing.<ul><li>Use the historical transaction dataset with attributes : Amount, Time, Location, User behavior,....</li><li>Cleaning the missing values, remove the duplicate, handle the outliers, remove the unexpected value</li></ul></li><li>Exploratory Data Analysis (EDA)<ul><li>Understand the normal and abnormal transaction patterns</li><li>Identify the useful data (features selection) by using correlation.</li></ul></li><li>Machine Learning Model.<ul><li>Use classification models such as Random Forest, XGBoost, or Neural Networks.</li><li>Also explore unsupervised techniques like Isolation Forest or Autoencoders.</li></ul></li><li>Evaluation.<ul><li>Focus on recall and precision (especially minimizing false negatives).</li><li>Use ROC curve, F1-score, confusion matrix.</li></ul></li><li>Deployment.<ul><li>Create a dashboard for monitoring.</li><li>Set up alerts for high-risk transactions.</li></ul></li></ol>

1. Objective :
Develop a machine learning model to detect suspicious transactions in real time and help prevent fraud in the financial system.

2. Why this case study :
Fraud detection is a critical issue for financial institutions. By using data science, we can automate the identification of unusual behavior and reduce losses. This case study reflects real-world challenges and demonstrates how data science creates business value.

Irenee Murhula

About me

Education

Strathmore University

Fundi Maendeleo Technical Institute

Portfolio

Fraud Detection.

Individual Project

Talents

Recruiters

About