E-commerce sites transfer a lot of money. This can lead to significant risks of fraudulent activities, such as the use of stolen credit cards, money laundering, etc.
Fortunately, Machine Learning can help us identify these fraudulent activities. All of the websites where you enter your payment information have a team that manages the risk of fraud through ML.
The goal of this challenge is to build a model that allows you to predict the probability of a fraudulent transaction.
Company X does E-commerce and sells handmade clothing. Your goal is to build a model that can predict whether buying a piece of clothing should be considered a fraudulent transaction or not.
Here's exactly what you need to do:
-
For each user, determine the country of origin from their IP address
-
Build a model that can predict whether the activity is fraudulent or not. Also explain your choices / assumptions in terms of optimizing false positives and false negatives
-
Your boss would like to understand your model because he is worried about using a black box model. How would you explain it from a user point, not mathematical. For example, which users can be classified as risked?
-
Suppose you can use your model live to make its prediction in real time. From a Product perspective, how would you use it? How could you think of the user experience with this product?
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent