Fraud Detection Case Study

Due Monday by 3am

Points 100

Submitting a website url

Start Assignment

OBJECTIVE: The objective of this assignment is to develop a machine learning system prototype to

detect fraudulent transactions. The ultimate goal is to give you practical experience in handling messy,

real-world data, designing, implementing, evaluating, and deploying all the components surrounding the

Fraud Detection System.

Deliverable A: Docker-enabled Software Package

Submit this deliverable through your provisioned repository on GitHub. Minimally, your system must

address the following requirements:

R1: The system should improve from the previous performance of the model.

R2: The system should be able to predict if a given transaction is legitimate or fraudulent.

R3: The system should allow administrators to generate a new dataset for training from the available

data sources.

R4: The system should allow administrators to select from a catalog of pre-trained models.

R5: The system should allow administrators to audit the system's performance.

Your code must be clean and organized for easy readability. Provide a Readme.md that provides user

instructions on how to run the software and demonstrates how the system meets the set of

requirements.

Deliverable B: Written Report

Submit a report as a markdown file (securebank/System_Report.md). The report must include design

information on:

System Design:

How your system design meets the requirements gathered.

Description of the unique functionalities of your developed modules.

Use accompanying diagrams to describe your system components and processes.

Data, Data Pipelines, and Model:

Description of the data and significant patterns you see that influence your design.

Explanation of data pipelines.

Description of the inputs and outputs of the model.

2024/9/19 20:36Fraud Detection Case Study

https://jhu.instructure.com/courses/82966/assignments/878719?return_to=https%3A%2F%2Fjhu.instructure.com%2Fcalendar%23view_name%3Dmo...1/2

Total Points: 100

SecureBank Case Study

CriteriaRatingsPts

30 pts

10 pts

10 pts

10 pts

30 pts

10 pts

Metrics Definition:

Detailed description of your offline and online metrics and their purpose.

Analysis of System Parameters and Configurations:

Feature Selection

Dataset Design.

Model Evaluation and Selection.

Post-deployment Policies:

Monitoring and maintenance plan.

Fault mitigation strategies.

SUBMISSION: You will need to check in all your code needed to run the system and provide the link to

your report (securebank/System_Report.md).

Docker-enabled Software Package

Implementation runs properly, and code is easily readable.

System Design

Data, Data Pipelines, and Model

Metrics Definition and Selection

Analysis of System Parameters and Configurations

Submission analyzes at least three significant design decisions and provides well-supported

evidence.

Post-deployment Policies

2024/9/19 20:36Fraud Detection Case Study

https://jhu.instructure.com/courses/82966/assignments/878719?return_to=https%3A%2F%2Fjhu.instructure.com%2Fcalendar%23view_name%3Dmo...2/2