Next, to calculate Z-scores, create a calculated field that subtracts each data point by the global average we just calculated, and divide the difference by the global standard deviation calculated. We’ll use them to get global averages and standard deviation, which are part of calculating Z-scores.Ĭreate a calculated field for the global average.Ĭreate a calculated field for the global standard deviation. We’ll be using WINDOW formulas, which perform an aggregation based on what is being displayed in the pane or window. Step 1: Get a Global Average and Standard Deviation This is actually really nice because Tableau will automatically read the data and set the center of the color scheme to zero, which now represents the average of all our data. This may sound confusing, but the results will be easier to interpret the mean of our data will be zero, leaving all negative values as below the mean and all positive values above the mean. Again, without diving into statistics, Z-scores represent how far away each data point is from the mean by using standard deviation. This option is slightly more involved, but worth it. Log transformation results: Option 2: Color by Z-scoresĪ more robust and interpretable solution would be to use Z-scores. However, it does improve our ability to evaluate the performance of regions in relation to others. We should also keep in mind that these transformations also change how we should interpret the colors: a value of 280.5 is no longer $280.50 and is more difficult to explain. Also, we were a little lucky that we didn’t have negative values or zeros in our dataset, which can present problems for square root and log transformations. We’ve reduced the influence of the outliers, but the center is still just a value in the middle of our minimum and maximum values. We can see that the square root transformation turned out better in our example, but we’re still faced with the same overall problem. In the new Tableau 9 it’s made even easier by just editing the calculation in the shelf: Without getting too deep into statistics, taking a square root of the data or log transforming the data are easy ways to stabilize data sets and reduce the effect of outliers.Ĭreate a calculated field that takes the square root or log of the metric you are using. Option 1 – Transform the Data Using Square Root or Log A better solution is one that will change with the data. However, if there are any updates to our data, the values may no longer be valid and we’ll have to re-guess each time. We can alter the start, end, and center, but we’d just be plugging in guesses as to what would look better, which might work for a one-time view. However, with our sales data, there are no negative values, so our data becomes centered around a value that is halfway between our minimum and maximum value, as you can see below: When there are positive and negative values present in the data, such as there would be in profit data, Tableau chooses zero, which works great in most cases. What Tableau automatically does is look for a logical center. To overcome this we’ve come up with a few simple workarounds to help in these situations and make the color feature more meaningful. The Superstore map comes colored by profit ratio, but what if this isn’t available? What if we want to compare by number of transactions or just raw profit or sales? Here’s what a switch to coloring by the Sales metric looks like:ĭue to the data being clustered on the low end of the range, this particular visual is not super helpful. Colors are very vulnerable to outliers, with these extreme values pushing all the color in one direction and thus making comparison significantly more difficult. The default color feature works well with categories and grouping (e.g., coloring by region), but coloring by a measure shows some vulnerability with default options. Maps in Tableau can be manipulated in a plethora of ways, but the simplest and most intuitive way to customize a map is by altering the color and size of the nodes.
Just perform a quick image search for “Tableau” and you’ll find pages of vividly colored maps-even the pre-made workbooks show off how cool the maps feature can be. In case you missed it, here is part four.
#Tabulo map series#
This is the fifth and final blog post in a multi-part series on Tableau.
0 kommentar(er)
