🔥 ChatGPT Tips Tuesday #22: Let's try multiple files with Code Interpreter!

💡 ChatGPT Tips for the day

Based on popular demand from last week’s poll, we will test uploading multiple files with Code Interpreter.

I came up with this testing idea as this is one of the new things released in ChatGPT early this August. See the tweet about this below:

For the test, I am thinking to test the following:

1. Join two CSV files to run data analysis

2. Combine two images together

For the first test, I manually created two CSV files. They contain some sales data in 2023 and basically, the month column is where they can be joined together for the analysis. The following are the dummy CSV files I have created:

In terms of the prompt, I would like to get ChatGPT to do the joining of the data to see the total sales I got from both items and highlight the highest month where I have got the biggest amount of sales. Lastly, I would ask ChatGPT to suggest some reason why the particular month has the highest amount of sales. I know I made up the data, so I am curious about the reason that ChatGPT would suggest.

The following is my prompt for this test:

The uploaded files are the sales data of apple and pear that I sold in my store in 2023. Please go through the files and do the following:
- Create a combined CSV file that show me the total sales for my store in 2023
- Create a data visualisation to present the total sales for my store in 2023 month by month and highlight the month that I have got the highest amount of sales
- Suggest reason why I have got the highest amount of sales in that particular month

The result I have got from Code Interpreter is amazing. It has successfully accomplished all the goals I have set in that prompt. The following is the data visualisation and the reasons that ChatGPT has suggested for the highest month.

If you are curious about how ChatGPT got to this result, feel free to check it here: https://chat.openai.com/share/4a984b46-13f3-4bdf-9a69-aa8a7a634f28. You can check how they get the results for each step using the Show Work function:

Alright, so Code Interpreter can do a good job of data analysis. But what about manipulating some images?

For the second test, I have two images where one is intended for the background and the other one is the main object to be included.

This image for the background:

The image for the main object:

For the test, I did try some different prompts but the results were not what I want. It did try to process the image, i.e. extracting the dog, but the result is actually not really good. Check it out:

In addition, I can’t verify if the images were combined as I can’t download the image:

So yeah, I would say Code Interpreter failed this task. Let’s see if the future version can do a better job. You can see my 3rd try with Code Interpreter here: https://chat.openai.com/share/8db76698-039e-4fe3-bb6e-a146b839aa7b

In summary, I think Code Interpreter becomes an even better tool for data analysis given the multiple files upload function. It did have some potential for image manipulation but I think it still needs a bit more improvement. If you have some other ideas to test the Code Interpreter, please let me know.

🖼️ Image of the day

🤣