credit_card_fraud_detection

Real time credit card fraud detection

1. Run with docker

ll

2. Testing Locally

To test the app locally, you first need to download and install Kafka and Cassandra.

2.1 Setup kafka

2.1.1 download kafka

Download Kafka from the following link:

https://kafka.apache.org/downloads

After downloading, navigate to the Kafka folder. Ensure the bin folder is in the current directory.

2.1.1 Start Zookeeper

Run the following command:

bin/zookeeper-server-start.sh config/zookeeper.properties

2.1.2 Start kafka server

Run the following command:

bin/kafka-server-start.sh config/server.properties

2.1.3 Creating a kafka topic

Run the following command:

kafka-topics.sh --create --topic transactions --bootstrap-server localhost:9092 --partitions 1 --replication-factor 1

2.2 Running Cassandra

2.2.1 Download Cassandra

Download Cassandra from the following link:

https://cassandra.apache.org/_/download.html

After installing, access the Cassandra shell:

cqlsh

2.2.2 Create KEYSAPCE

Run the following commands:

CREATE KEYSPACE fraud_detector WITH REPLICATION = {'class': 'SimpleStrategy', 'replication_factor': 1};
USE fraud_detector ;

2.2.3 Create a table

Run the following commands:

CREATE TABLE transactions (
    transaction_id INT PRIMARY KEY,
    status TEXT
);

The status indicates whether the transaction is fraudulent or not.

2.3 Setting up Front end

I am using react for the front end, so you should have node installed in you machine. You can do that by doing the following :

https://nodejs.org/en/download/prebuilt-installer/current

After downloading, ensure Node.js is added to your PATH. Verify the installation by running:

npm -v

Navigate to the frontend folder and install the required packages:

npm install 

Run the development server:

npm start

The React app should run on http://localhost:3000 by default.

2.4 Setting up the backend.

Navigate to the backend folder.

2.4.1 create a conda env

Run the following command:

conda create -n credit_fraud python=3.11 -y

2.4.2 install all requirements

Run the following command:

pip install -r requirements.txt

2.4.3 create .env

Create a .env file in the main project folder (credit_card_fraud_detection) with the following content:

PYTHONPATH="/path_to_the_main_folder/credit_card_fraud_detection"

CASSANDRA_HOST=localhost

CASSANDRA_KEYSPACE=fraud_detector

KAFKA_BROKER=localhost:9092

2.4.4 Run the backend

Ensure Kafka is running (refer to previous steps).

Export the PYTHONPATH before running the app (replace path_to_main_directory with the actual path):

export PYTHONPATH="path_to_main_directory/credit_card_fraud_detection"

Run the backend:

uvicorn cc_fraud_detector.main:app --reload --host 0.0.0.0 --port 8000

Now you also need to run the Kafka consumer so it consumes the transactions and triggers prediction.

You need to open another terminal, navigate to the backend folder, and run the following command:

python -m cc_fraud_detector.api.services.kafka_consumer

If you face any errors, run the export command again:

export PYTHONPATH="path_to_main_directory/credit_card_fraud_detection"

Now you can use the web app to simulate a transaction and wait for the prediction.