Detect Credit Card Fraud in Just Few Minutes

Author(s): Sara M.

Originally published on Towards AI.

In this article I want to show you how a complex analysis in a critical sector like banking could be done in a short time and without programming expertise.

So, for the demo, I picked the Pranjal Saxena article that proposes a solution using Machine Learning & Python to do the analysis.

The point is to try to do the analysis with only ChatGPT4.

The dataset is the same as used in the article.

Let’s begin.

“Data Processing & Understanding” Step

Without importing any library and without writing any line of code, we will just ask ChatGPT to give us the information

And his answer after analyzing the dataset we give him is:

The result are the same as Pranjal Saxena result.

Isn’t it amazing ?

We got the result with 0 code, no environment needed, not a single library…

Next step, is Data Cleaning. We will ask him to remove duplicates if found.

The same result was found again.

What is interesting is that you can see the code used !

“Train & Test Split & Model Building” Step

I have asked him without going into details to check the accuracy of “decision tree model” :

Here again, we got the same result (99,92% ) as Pranjal Saxena result.

and about the “F1-score” check:

We can go further and check the other models but it’s not the purpose.

ChatGPT doesn’t need you to tell him how to Analyze

Until now, the analysis done was guided by us.

What if we tell him to do his own analysis ?! (I upload a test file with 100 000 credit cards)

And the surprise that he proposes a detailed approach to analyze the Data which I found very complete.

Here is his answer:

Some results from the “Data Inspection” Step:

Some results from the “Exploratory Data Analysis” Step:

Conclusion

This demonstration underlines the transformative potential of AI in Data Analysis. ChatGPT-4’s ability to replicate expert-level analysis without requiring coding skills or specific environment setups signifies a shift in how data-driven decisions can be approached. The next article will be about automating the AI analysis process with a data pipeline using AWS services.

If you have any questions, feedback, or would like to share your experiences, please feel free to reach out in the comments section.

Clap my article 50 times U+1F44F, that will really help me out and boost this article to others U+270DU+1F3FBU+2764️. Follow me on Medium to get my latest article.

Thank you U+1FAF6!

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI