Tutorial   #promptengineering #chatgpt #dataanalysis  

This is my takeaway of a Youtube video tutorial of using ChatGPT to perform #DataAnalysis.

Skills you need to learn

In order for ChatGPT to perform a specialist role in a certain area, as part of #PromptEngineering skillset, you need to learn how to customize #ChatGPT to assume your role and achieve your goal in your specialist field. That means before using ChatGPT, you need to provide answers to the following two questions in Custom instructions for ChatGPT, a feature launched recently in July 2023.

Question 1: What would you like ChatGPT to know about you to provide better responses?

Example Answer:

• Profession/Role: Yoga instructor
• Key Responsibilities: Leading yoga classes, maintaining a safe and tranquil environment, guiding students in their practices
• Knowledge or Expertise: Yoga philosophy, various yoga poses and sequences
• Typical Challenges: Adapting teaching methods to suit different skill levels, maintaining a peaceful class environment
• Current Projects: Virtual yoga classes

Question 2: How would you like ChatGPS to respond?

Example Answer:

• Tone and Formality: Calm, peaceful and instructive
• Level of Detail: Detailed pose descriptions and breathing techniques.
• Preferred References: Yoga traditions, wellness literature
• Examples or Analogies: Well-known yoga sequences, meditation techniques
• Avoidance of Ambiguity: Clear and direct yoga guidance
• Resource links: Yoga platforms, meditation apps

What can ChatGPT do for you?

Other than allowing you to upload data file, I observe the following data analysis capabilities from ChatGPT in the demo

• Generate well-worded data dictionary for each column if column name is self-explanatory. For example: for Purchased column name, it came up with A binary variable indicating whether the user made a purchase (1) or not (0)
• Remind you detail list of tasks given a general intent. For example: given a general intent of "cleaning data", it suggests a list of action items including Check for Missing Values, Data Type Verification, Outliers etc.
• Carry out one or multiple tasks. For example: if instructed, ChatGPT will perform the data cleaning tasks in previous step on provided data.
• Give suggestions to next step after completing a task.

Overall Takeaway

As a data engineer for over 10 years, features of ChatGPT that I see useful at this point are:

• help me write technical documentations or tutorials.
• generate well-worded column-level data dictionary if column names are self-explanatory enough.
• give me suggestions of tasks given a certain intent or goal.

At the same time, I observe the following limitations as of now:

• Cannot use its analysis result now due to lack of ways to verify its results or prevent it from hallucinating.
• Cannot use it to do innovative tasks or tasks that are not well-documented.
• Cannot use it to analyze large amount of data due to limitation of compute power.