1 of 7

Data Analysis using Generative AI - ChatLoS-DTA - A Study Using the Domestic Traffic Ads Legacy of Slavery Dataset

Spelman College Datathon, Nov 1, 2024

1

2 of 7

Domestic Traffic Ads Dataset

2

Newspaper advertisements placed by slave traders to buy or sell enslaved throughout the state of Maryland between 1824 - 1864.

3 of 7

Domestic Traffic Ads Dataset - Maryland State Archives Online Database

3

4 of 7

Domestic Traffic Ads Dataset - Sample used for this exercise

4

A sample of 764 from the 2800+ DTA dataset records were exported in a CSV format (screenshot below)

5 of 7

5

Using Generative AI to perform Data Analysis on the sample DTA dataset

  • Introduction of ChatGPT (Generative AI LLM based chatbot)
    • Generative Pre-Trained Transformer (GPT)
    • Powered by Pre-Trained Large Language Model (LLM)
    • Understands natural language statements and converse with users by generating human-like text and images
  • A latest advanced feature is to perform data analysis
    • Don’t need to know the column names of the underlying data
    • Converts natural language questions into python queries
    • Can create interactive visualizations
    • Quick setup

6 of 7

6

Using Generative AI to perform Data Analysis on the sample DTA dataset

  • Creating a custom GPT, ChatLoS-DTA to perform data analysis on the Domestic Traffic Ads (DTA) dataset by uploading it.
  • Let’s chat with this custom GPT!!
    • How to access this custom ChatLoS-DTA using a free account sign up1
    • A list of questions that could be asked to the ChatLoS-DTA here
    • Or you can share the questions in the chat so I can share and show the results.

1 - There are usage limitations on ChatGPT with free account for data analysis conversations. 4 messages per user per day

7 of 7

Questions?