Exercise: Finding Patterns in Support Data

The support ticket dataset is different from revenue data. Revenue is aggregate and trends over time. Support tickets are individual events with multiple categorical and numerical dimensions. This changes the visualization strategy entirely. You are looking for distributions, correlations, and segments — not trends and totals.

Here is a support ticket dataset. Analyze it and recommend visualizations for these questions:

1. Is there a relationship between ticket priority and resolution time? What chart shows this best?
2. Do escalated tickets have lower satisfaction scores? How should I visualize this comparison?
3. Which agent handles the most escalations, and is their resolution time better or worse than average?
4. What is the distribution of resolution times — is it normal, skewed, or bimodal?

For each recommendation, explain why that chart type works and what alternatives would fail.

[Paste support-tickets.csv here]

In Claude, this prompt will produce both analytical reasoning and a rendered Artifact with the recommended charts. In ChatGPT, you will get a Python script to run. Either way, the key is forcing the AI to reason about data relationships before choosing chart types. A scatter plot of priority vs. resolution time only works if priority is treated as ordinal — the AI needs to map Critical/High/Medium/Low to a sensible axis. A box plot grouped by priority would show distributions more clearly. The AI should recommend it.

Upload support-tickets.csv to ChatGPT, then try these in sequence:

Prompt 1: "Create a Python script that generates a box plot of resolution hours grouped by priority level. Order the priorities: Critical, High, Medium, Low. Dark theme, cyan accent. Title: Resolution Time by Priority."

Prompt 2: "Add a second chart: scatter plot with resolution hours on the x-axis and satisfaction score on the y-axis. Color points by whether they were escalated (red for Yes, cyan for No). Add a trend line for each group."

Prompt 3: "Add a third chart: grouped bar chart showing average resolution time by agent, split by escalated vs. non-escalated. Add error bars showing standard deviation."

Prompt 4: "Combine all three into a dashboard layout (2x2 grid, leave bottom-right empty or add a summary stats table). Title: Support Operations Q1 2025. Save as a single PNG at 200 DPI."