Build Fraud Detection Model and Detector

info

• Follow the steps mentioned below.
• Total time taken for this task: 2 hours .
• Pre-requisites: None

Tidbits

• AWS Fraud Detection service is not natively supported by Amorphic.
• We are going to access it using an ML notebook that needs specific access from the AWS console.
• An ML notebook is created and shared with you. It has all the necessary access for building the model.
• Do not try creating a new ML notebook. You will run into access-related issues.

Download fraud detection notebook

• Go to the Amorphic Dataset: fraud_detection_resources. Use the navigator to quickly go to the Dataset page. Press ctrl two times successively and type the dataset name fraud_detection_resources.
• Click on the files tab.
• Click on three dots ⋮ in front of the BuildFraudDetectorModel-YourUserMame.ipynb file and click on Download File.
• Rename the file to replace YourUserName with <your_userid> in BuildFraudDetectorModel-YourUserMame.ipynb.

Open an ML notebook

• Click on 'MACHINE LEARNING' --> 'Notebooks' from left navigation-bar.
• You will see 'ml-notebook-ankamv' in the list.
• Click on 'View Details'.
• If the 'Notebook Status' is not 'InService', follow the below instructions.
  • Click on ✏️ icon to edit it. Change 'Auto Termination Time' to next 6 hours and save changes.
  • Click on ▶️ icon at the top right corner to start the notebook.
  • Click the 🔄 icon to refresh the status.
  • Once the status turns to 'InService', you will see a link Notebook URL Link as shown below.
  • Wait for sometime and click on the 'Notebook URL'. try after sometime if it is asking you to sign-in to AWS console.
• If the 'Notebook Status' is already 'InService', click on the 'Notebook URL'.

Upload and prepare notebook

• Click on Upload button in the jupyter notebook as shown below.

• Select the above BuildFraudDetectorModel-<your-userid>.ipynb and then click 'upload'.
• Click BuildFraudDetectorModel-<your-userid>.ipynb to open it.
• Change your username as highlited in below picture.

Build Model

• Run each cell one by one to create and test the model.
• This notebook creates the following items on AWS backend.
  • Model
  • Predictor
  • Event
  • Entity
  • Outcomes
  • Labels
  • Variables
• This notebook predicts a sample data to test the model. Check MODEL_SCORES, OUTCOMES, STATUS variables at the end of the notebook.

Congratulations!!!

You've learned how to build a fraud detection model on Amorphic. Now, proceed to 'Ingest stream events' task.