Logo Lanfrica

lloydeila/mpesa-fraud-detection-system

Domaine:

digital infrastructure

Type de record:

software
Créateur:
llo
HĂ´te:
Mpesa mobile application built using MIT App Inventor # Mpesa Fraud Detection System ## 🔍 Overview This project is a mobile-based fraud detection system built using **MIT App Inventor**. It simulates fraud detection in mobile money transactions using a **hybrid approach combining rule-based detection and a Naive Bayes probabilistic model (conceptual/extended).** The system also uses **TinyDB for local data storage** to record and analyze transaction history. --- ## 🚨 Problem Statement Mobile money platforms are vulnerable to fraudulent activity such as: - Suspicious transaction behavior - Repeated unauthorized transfers - Social engineering-based fraud patterns This system demonstrates how fraud can be detected using a combination of **keyword-based rules and probability-based classification concepts**. --- ## ⚙️ Features ### 🔹 Rule-Based Detection (Keyword Pattern Matching) The system uses predefined **fraud-related and genuine keywords** to evaluate transactions. - Fraud indicators include suspicious or high-risk keywords - Genuine keywords indicate normal or trusted behavior - Transactions are analyzed based on the presence of these words If fraud-related keywords are detected, the system flags the transaction as **potentially suspicious**. --- ### 🔹 Naive Bayes Probabilistic Model (Conceptual Layer) The system is extended with a **Naive Bayes machine learning concept** to estimate fraud probability based on transaction patterns. This model evaluates the likelihood of fraud using: - Transaction behavior patterns - Frequency of activity - Keyword occurrence probability - Historical classification trends :contentReference[oaicite:0]{index=0} ### Purpose: - Assign probability scores to transactions - Improve decision-making beyond simple keyword rules - Demonstrate machine learning enhancement potential --- ### 🔹 Data Storage (TinyDB) - Stores transaction records locally on the device - Maintains history for comparison and analysis - Supports rule-based and conceptual ML evaluation --- ## 🛠️ Tools Us …

Languages