Charting tips 002: Consider the task at hand

Suppose you are sharing a list of orders with some co-workers. One of them wants to see the higher sales orders

[list]. Another one wants to know how much was exported to France [fusion_old_table]. The next one needs the average items per order [descriptive statistics]. You want to see the growth trend for several products [chart]. Only one of the tasks requires a chart. So, be sure to understand the nature of the task and then select the right tool.

8 thoughts on “Charting tips 002: Consider the task at hand

  1. I guess next tip will be “Hey guys! For a seeing of trend we don’t need to draw a chart. One figure is enough. Growth rate :)”

  2. grossu

    Thanks for the tip 🙂

    But growth trend and growth rate are not exactly the same thing… and you can’t have just one figure to tell you the growth rate of *several* products (one by one).

  3. The problem is that some people try to create a single display element to show all of these disparate measures. I don’t know if this is a relic from the days when a PC had miniscule RAM and used only 5.25″ floppies, but now that we routinely have 1+ GB of RAM and 200 GB of hard drive storage, it makes sense to use as many sheets and tables as are needed to tell the story slearly.

    Most effective is four separate display elements for the four bits of information sought by the information consumers. They are all based on the same data, or at least subsets of the same data. The original data should be held away from the consumers, and accessed only through the separate displays.

    A corollary to this is my suggestion to people to make a separate worksheet for each use of the information, e.g., one sheet formatted for printed display, another for on-screen display, another for each particular arrangement is required for each chart, and perhaps more for each arrangement needed as a precursor to a particular statistical analysis. Worksheets are cheap.

  4. If these are the requirements to create a dashboard (?), then I think it’s time to re-evaluate what your co-workers really need.

    Add context to every asked report, such as a possibility to compare highest sales orders on every month, exports to France compared to other countries per year and month, etc. Try to find common denominators, then build a dashboard that everybody wants. It’s the job you must do, because you are a BI analyst. 🙂

  5. Janne: these new (and hopefully daily…) charting tips are not targeted specifically at dashboard design. With these first two tips I just want to underline the fact that sometimes a chart is not the best tool and that the analyst should consider other options.

    There is a fine balance between finding common denominators that could end in a one-size-fits-all approach that no one really likes and custom-tailored solutions impossible to manage. Yes, a BI analyst is also paid to find that balance…

  6. Jon: I usually don’t take user’s requests at face value. I try to solve their problem and give them a little more than they expect. Most of the time this means merging similar requests into a single display and make them interact with the tool to get what they want.

    So, I agree with you regarding the use of separate worksheets, but how to separate them is not always easy.

  7. Charts or something else… with my seven-month-experience in the largest BI company of Finland, I must say charts are 5-10% minority (though a growing minority). I have seen a project where my colleagues worked 2+ years to create a reporting solution with no charts at all (and the tool they use has adequate charting options). The customer is totally satisfied with lists and tables as long as they can export numbers into Excel and create their own charts.

    Sad to say, the common knowledge about data visualization is not yet high in Finland.

    How about other countries? Is the table:chart ratio 10:1 or 1:1?

  8. By the way, why don’t you create a scatter plot? Here’s a parallel between your old/young XY and the task at hand:

    Countries in old/young XY = various products sold, or perhaps a matrix formed of products and countries where the products are exported = in this case the number of symbols in XY will be number-of-products * number-of-countries

    Variables in X and Y axis = various measures such as total sales, highest sales orders, average order in quantity, price, and money, additional information such as the market size and share of a product in individual countries, costs of products if available, etc (some measures need logarithmic scale such as total sales)

    Symbol size = total sales or market size of product in the country

    Symbol color = changes when user selects some product or country

    Animation = time (years and months)

    Benefit = the XY chart with various measure combinations available for the X and Y axes would contain much more information than any other chart

    I’m currently working on a (long) project where I’m planning to use this kind of parallelisms for data visualization.

Comments are closed.