DETECTIKA

NSEIT brings the power of fraud detection directly to your Snowflake account. Using Streamlit and Snowpark, the application enables you to detect fraudulent credit card transactions with no need to transfer files or share data outside of your Snowflake Data Cloud.

Using the fraud detection model built by NSEIT "DETECTIKA" which is trained over 100k publicly available credit card transactions, the client can detect fraud for millions of transactions in one go.

For each input record provided the output will return a fraud probability score which will determine whether the recorded transaction is fraud or legitimate. Higher the score, higher the probability that the transaction is fraud.

Expected workflow for client specific outputs:

1. Customer sends email to NSEIT at: [e-mail] with the following information.
-type of transactions
-transaction table metadata ( for identifying customer specific features )

2. NSEIT will access client metadata ONLY to prepare the logic/sql queries, so that client can make use of the existing pre-trained model "DETECTIKA". (Estimated time to deliver is 2 business days)

3. NSEIT will notify client when they will be able to run the application in their account to detect fraudulent transactions.

4. NSEIT's team to install DETECTIKA in their snowflake account if required.

搜集汇总

背景与挑战

背景概述

DETECTIKA是部署在Snowflake平台上的信用卡交易欺诈检测系统，基于10万条公开交易数据预训练模型，可批量分析交易并提供欺诈概率评分。客户需提供交易类型和元数据，由NSEIT团队在2个工作日内配置后即可在客户账户中运行检测。