Credit Card Fraud Detection using Spark ML

Course - CPSC 531 Advanced Database Management

Developers -

• Sarita Joshi
• Hritika Phule

Overview

This project demonstrates a robust implementation of a real-time credit card fraud detection system using Apache Spark ML, integrated with Apache Kafka, Zookeeper, and Cassandra for data streaming, processing, and storage, respectively. A live dashboard for reporting and analytics is created using Python Streamlit.

Technologies Used

Each component plays a crucial role in the system:

• Data Generation: Python Script
• Data Streaming: Apache Kafka + Zookeeper
• Model Training: PySpark
• Data Processing: Spark and Python Script
• Data Storage: Cassandra
• Data Reporting and Analytics: Streamlit Python Live Dashboard

Data

Dataset link - https://www.kaggle.com/datasets/kartik2112/fraud-detection

Dataset generator - https://github.com/namebrandon/Sparkov_Data_Generation/

The dataset was generated using Sparkov Data Generation | Github tool created by Brandon Harris.
The simulator has certain pre-defined list of merchants, customers and transaction categories. And then using a python library called "faker", and with the number of customers, merchants that you mention during simulation, an intermediate list is created.

Datset size- 1GB

Info

#	Column	Non-Null Count	Dtype
0	trans_date_trans_time	14814000 non-null	object
1	cc_num	14814000 non-null	int64
2	merchant	14814000 non-null	object
3	category	14814000 non-null	object
4	amt	14814000 non-null	float64
5	first	14814000 non-null	object
6	last	14814000 non-null	object
7	gender	14814000 non-null	object
8	street	14814000 non-null	object
9	city	14814000 non-null	object
10	state	14814000 non-null	object
11	zip	14814000 non-null	int64
12	lat	14814000 non-null	float64
13	long	14814000 non-null	float64
14	city_pop	14814000 non-null	int64
15	job	14814000 non-null	object
16	dob	14814000 non-null	object
17	trans_num	14814000 non-null	object
18	unix_time	14814000 non-null	int64
19	merch_lat	14814000 non-null	float64
20	merch_long	14814000 non-null	float64
21	is_fraud	14814000 non-null	int64

System Architecture

Installation

Ensure all dependencies and required technologies are installed. Refer to official documentation for Apache Spark, Kafka, Cassandra, and Python for detailed installation instructions.

To get started with this project, you will need to set up your environment with Apache Spark, Apache Kafka, Zookeeper, and Cassandra. Below are detailed steps and links to resources for installing and configuring each component.

Prerequisites

• Java 8
• Python 3.11

Apache Spark

1. Download Apache Spark:

   • Download the latest version of Apache Spark from Apache Spark Downloads.
   • Choose a package type that supports Hadoop.

2. Installation:

   • Unzip the downloaded Spark package to a directory of your choice, e.g., C:\spark.

3. Environment Variables:

   • Add C:\spark\bin to the PATH environment variable.
   • Set SPARK_HOME to C:\spark.

Apache Kafka + Zookeeper

Apache Kafka requires Zookeeper for coordination.

1. Download Apache Kafka:

   • Kafka can be downloaded from Kafka Downloads. Select the binary files that include the Scala version you prefer.

2. Installation:

   • Extract the Kafka binaries to your preferred location, e.g., C:\kafka.

3. Configuration:

   • Edit the Kafka configuration to set up data directories. Detailed configuration steps can be found in the Kafka documentation.

4. Environment Variables:

   • Add C:\kafka\bin to the PATH environment variable.

Cassandra

1. Download Apache Cassandra:

   • Download it from Apache Cassandra Downloads.
   • Ensure that the installed version contains .bat files for windows.

2. Installation:

   • Extract Cassandra into a directory, e.g., C:\cassandra.

3. Environment Setup:

   • Set CASSANDRA_HOME to C:\cassandra.
   • Add C:\cassandra\bin to the PATH environment variable.

Workflow

The development and implementation of this project follow these steps:

1. Start environment

Start zookeeper, kafka and cassandra using below commands.

• Start zookeeper C:/kafka/kafka_2.12-3.7.0/bin/windows/zookeeper-server-start.bat C:/kafka/kafka_2.12-3.7.0/config/zookeeper.

• Start kafka C:/kafka/kafka_2.12-3.7.0/bin/windows/kafka-server-start.bat C:/kafka/kafka_2.12-3.7.0/config/server.properties

• Start cassandra First navingate to the bin directory in cassandra and write below commands. cd C:/cassandra/bin

cassandra

2. Data Cleaning and Preprocessing

Input data is cleaned and preprocessed to ensure quality and compatibility for model training, including feature encoding and normalization.

3. Model Training and Evaluation

A Random Forest model is trained using Spark ML, evaluated for accuracy, and the model is saved locally.

4. Data Producer Code

A Python script generates random credit card transactions and publishes them to a Kafka topic.

5. Data Processing

The trained model processes transactions from the Kafka queue in real-time to predict fraud, and results are stored in Cassandra.

6. Reporting

A Streamlit dashboard displays real-time data, graphs, and analytics, pulling data directly from Cassandra.

Challenges

1. System Compatibility issues
2. Synthetic Data generation
3. Cassandra setup - Memory allocation issues