Python hackathon

During the hackthon we will build a web app that does fraud detection on (fake) transactional data stored in a Postgress database.

Prerequisites

• Python 3 virtual environment with the following packages:

    aniso8601==1.1.0
    Flask==0.10.1
    Flask-Compress==1.3.0
    Flask-REST==1.1
    Flask-RESTful==0.3.5
    itsdangerous==0.24
    Jinja2==2.8
    MarkupSafe==0.23
    psycopg2==2.6.1
    python-dateutil==2.4.2
    pytz==2015.7
    six==1.10.0
    Werkzeug==0.11.2
  

• Postgres version 9.5 or later

• your favourite IDE

Outline

During the handson you will:

• build a transactional data generator
• setup a postgres database for storing the transactional data
• build a flask app that provides api to the postrgres db, the app
  • will have several different API end points
  • allows inserting of data via the api
  • performs fraud detection on transactional data
• optional: write test cases for your code
• optional: package the your app

Step 1: Transactional data generator

Create a fake transacitonal data generator that creates data in the following format:

• account id; integer
• transaction id; integer
• timestamp; string or int (resolution ms)
• transaction value; decimal

Further requirements on the generator:

• the time difference between generated transactions should vary
• the generator should be able to store data in the db
• through the app api using post requests
• direct inserting into the db
• should be configurable thoughs some simple parameters; eg post_url, num_transations, etc

Step 2: Setup the Postgres db

Setup a database such that:

• your app is be able to access the db (create a user etc..)
• it contains a table that can store the transactional data

Step 3: Create the Flask app

We will built the app by further extending the provided skeleton.

The api should at least implement the following two end points:

• /is_fraud/<int:account>/ (GET): returns fraudulent events for the specified account
• /store_data (GET + POST): stores transactional data into the db

The fraud detection logic can be implemented in two ways:

• in the db by creating a UDF
• in python code by applying detection to data retrieved from the db

The decection logic itself should be based on a couple of simple rules (eg. amount < 0.9 average(amount))

Step 4 (optional): Write test cases

Create some nosetests for your developed python app.

Step 5 (optional): Package your app

Create a Python package for easy distribution and installation of your app.