Course: Data Communication in the Age of AI
Institution: University of Notre Dame
Faculty: Brandon Erlacher

Assignment: Identify interesting time-series patterns

a. Collect a small dataset (100 rows) that contains a date field, at least 3 measures and other categorical fields for variables of your interest (e.g., running data where each row would contain a value for each of the following columns: date/time, heart rate, temperature, activity name, pace, stride length, weather and V02 Max.) Make sure each row of data contains a date/time. Some interests that you could collect data include: screentime, email, music listening history, sensor data from wearables, sports related stats, visually observed data points over time, financial data (banking, credit card). Then provide a brief reflection to acknowledge Data Fidelity & Limitations (Humility) with your data:

• • Blind Spots: At least two analytical questions this dataset cannot answer.
  • Assumptions: Where you had to make clean-up choices (e.g., how you handled null values, outliers, or missing data segments).
  • Potential Biases: At least one systemic limitation or potential collection bias in the underlying pipeline.

b. Create visualizations in Tableau to demonstrate interesting insights from your data. Make sure to add the appropriate context (e.g., annotations, captions, reference lines, trendlines, quick-table calculations, forecasts) where necessary.

c. Evaluate relationships between measures in your data. Create at least two scatterplots and describe the regression conditions for each plot in an annotation or on a dashboard with a text box. Briefl y explain the relationship between the two measures and why or why not it is appropriate to use the explanatory variable as a predictor. Make sure to add a linear trend line to help with the evaluation.

d. For all the visuals you have created, include a caption that contains a Provenance & Attribution Statement (Gratitude). You must address the following:

• Infrastructure: Who built, maintained, or cleared this data for your use?
• Subjects: Whose actual lives, actions, or assets are represented by these rows of numbers?
• Tech Ecosystem: Acknowledge the broader open-source community, software engineering teams, or peers who provided the tools or troubleshooting support to bring this analysis to life.

Your visualizations will be evaluated based on the following criteria: creativity, analytical
depth and beauty/design.

e. Choose one of your time-series visualizations and use an AI of your choice to remake it outside of Tableau. Upload your data to the AI, then craft a prompt that will not only create the time-series, but include all the context as well (reference lines, trendlines, forecasting, annotations, legends, colors, etc.). Upload the output (e.g., a screenshot or a jpg/png) and your prompt/chat with the AI. Include a brief refl ection on your experience with using the AI to create the time-series instead of building it in Tableau. You may add this portion into your Tableau workbook by making a dashboard and bringing in text and image objects OR you can upload a separate PDF to Canvas.

f. Save your visualization(s) and data and submit on Canvas.