Asking Interesting Questions from Data

==2024-12-23


<% tp.file.cursor(4) %>


Theory

Good questions are one that spark a new curiosity towards a given subject. There are many types of questions to ask for a given dataset.

  1. Descriptive Questions: They describe the properties and metrics in the dataset.
  2. Diagnostic Questions: They provide answers to why certain phenomenon is occurring.
  3. Predictive Questions: They predict some feature of the data.
  4. Prescriptive Questions: They provide the decision making answers for an entity. Those decisions may be the actions to be performed to solve the given business problem.

Examples

  1. Sales Data

    Dataset Description: A dataset that includes information such as sales date, product name, sales amount, customer demographics, and region.

    Relevant Questions:

    • “What are the top-selling products in each region?”
    • “Can we predict which products will have the highest sales next year?”
    • “How can we adjust marketing efforts to boost sales in underperforming regions?”
  2. Social Media Data

    Dataset Description: Information about user engagement on social media, including likes, shares, comments, and post categories.

    Relevant Questions:

    • “Which type of posts generate the most engagement?”
    • “Can we forecast the engagement level of a new post based on past trends?”
    • “How should we tailor our content strategy to maximize user engagement?”

PTR