INFORMS Open Forum

  • 1.  Visualizing Insights: Seeking Recommendations 📊

    Posted 10-27-2023 17:42

    Greetings, Informs community!

    As I venture into the realm of data visualization, I'm curious about your approaches to selecting the most effective visualization methods for different types of data. Whether it's wrangling time series, working with categorical data, or navigating geospatial information, I'd greatly appreciate your insights and recommendations.

    Looking forward to hearing your thoughts!



    ------------------------------
    Shayan Farshid
    UC Davis
    San Francisco CA
    ------------------------------


  • 2.  RE: Visualizing Insights: Seeking Recommendations 📊

    Posted 10-28-2023 08:44

    That is a great question, but I fear no great answers!

    A lot depends on what you or your viewers want/need to learn from the visualization.  For example if you want to show there is a lot of variation in the data, you might use a line chart, but set the lower "Y" value of the chart at the lowest data point, and scale so the highest point is the top of the chart.  On the other hand, if you want to show the data is reasonably "flat", the set your lower point at 0.

    If you want to show two sets of data on the same chart, probably a mix of line and bar charts works.  Also Area charts with bar charts can be effective.

    Clusters and pseudo 3-d charts also have their place.  Even think about how your viewers are used to seeing data, what are they familiar with.  If you want to chock them, don't use that style of visualization.  If you want them to understand easily, use what they are familiar with.

    So, lots of options...



    ------------------------------
    Duncan Klett
    ------------------------------



  • 3.  RE: Visualizing Insights: Seeking Recommendations 📊

    Posted 10-30-2023 01:31

    Think of your dataviz in three categories:
    1. Professional comms with non-OR readers
    2. Intra-OR comms
    3. Just for fun

    In professional comms with non-OR readers:
    1a. Dimensions
    i. For Top management, Finance, & Accounting: 1 dimension
    ii. For Supply Chain, Manufacturing, Logistics: 2 dimensions
    iii. For analytics-aware audiences: 3 dimensions in column-and-line charts or bubble charts

    1b. People get confused by:
    i. Sort orders
    ii. Cumulative axes
    iii. Log scales

    1c. Thumb-rules:
    i. The "3D = column-and-line charts or bubble charts" rule is a cognitive boundary
    ii. Don't use pie charts

    FYI, once in my dataviz life I loaded 5 dimensions in one graph. It was for the geniuses in Dow R&D. A bubble chart naturally shows 3 dimensions (x, y, and bubble-radius). I used a color scale for the 4th, and labels for the 5th.



    ------------------------------
    Rahul Saxena
    FrogData.com
    ------------------------------