How to build a serverless real-time credit card fraud detection solution
HTML-код
- Опубликовано: 2 мар 2021
- More Smart Analytic Reference Patterns → goo.gle/3beq3j3
Code on Gitlab → goo.gle/3qa3Dng
Learn how to use serverless tools on Google Cloud to build a real time credit card fraud detection solution.
This is a step-by-step video that explores the credit card fraud detection pattern in this and helps walk you through the entire process of building such a system in your organization.
You will learn how to:
- Prepare the training data on BigQuery
- Train and evaluate the fraud detection model using BigQuery ML
- Build a streaming pipeline for online predictions on new transactions using Dataflow and AI Platform
- Set up fraud notifications using Pub/Sub
- Visualize using operational dashboards using Data Studio
Subscribe to Google Cloud Tech → goo.gle/GoogleCloudTech
product: Cloud - Data Analytics - BigQuery, Cloud - Data Analytics - PubSub, Cloud - Data Analytics - Dataflow, Cloud - Data Analytics - Google Data Studio; fullname: Polong Lin, Pavan Kumar Kattamuri, Eshan Tyagi; Наука
Can you help with the dataset used??
If I understand the presentation, the IS_FRAUD flag is said to be "between zero and one." But later it seems that the flag is set to EITHER zero OR one, which makes more sense. Boolean logic, not a range of fractional numbers.
Helpful patterns - Thanks for sharing!
Glad it was helpful!
@Rene Kohen yea, I have been using InstaFlixxer for since december myself :)
Not able to access the code on gitlab. Is it pubic repository?
Did not understand why Dataflow invoke model step is taking 22 minutes. Should not be on the fly?
This looks more like an ad for google clouds more than it actually being a useful module that can be used by different platforms
brother can you translate the video in hindi language