Tableau has some amazing built-in forecasting features. In this short lesson, we will get used in some domains more than others, but if it applies to you, it makes building a simple forecast easy. With each version of Tableau, the Tableau forecasting features get more advanced.
When is Tableau Forecasting Useful?
- Predicting sales trends and understanding spikes in demand
- Understanding seasonality
- Understanding how much inventory you need on hand at any given point in time
- For production and manufacturing planning on building and launching new products
- Predicting email response rates (though this would be even better if used in conjunction with Clustering)
- Understanding peak wait times in stores or lines, or peak usage times for mobile devices (to help in tuning cell towers)
- Predicting busy times at a hospital or call center by hour, day, or month to optimize staffing
- Essentially, any time you want to have a sense of what is coming in the near future
Tableau Forecasting Tutorial
- On a new sheet, drag Order Date to Columns and Sales to Rows.
- Change the Order Date to continuous by selecting the Month value in the second segment of date options. (i.e. Month May 2015)
- Right click on the chart and choose Forecast.
- You’ll now have a sales forecast for the next 12 months. It gave us a straight line, so let’s edit the forecast.
- You can edit the forecast by right clicking on it and choosing Forecast Options. Under Forecast Model choose “Custom.” Then we need to set Trend and Season to Multiplicative. Editing forecast options can get pretty technical, so do your research before showing the forecast to the CEO, just in case he throws you a curveball question.
- For more information on when to use Additive vs. Multiplicative, see here: http://support.minitab.com/en-us/minitab/17/topic-library/modeling-statistics/time-series/time-series-models/additive-and-multiplicative-models/
- But for the most part, you can be pretty comfortable using either. In this case, the multiplicative model was a little more accurate.
- A good way to test this is to use some data to create a forecast and then test it against data you’ve withheld. For example, using data from 2013, 2014, and 2015, build a forecast for each month of 2016. Compare the values you that the forecast provides vs. the actuals for the year. They should be fairly close. You can test this with additive models or multiplicative, or some combination to see which one gets you closest overall. Know that the further out you project, the more variation you will get.
- The great news is that Tableau will update the forecast each month, so you’ll always be pretty accurate a few months into the future.
- You can check accuracy with the formula (Forecast – Actual)/Forecast to determine percent variance.
Tableau forecasting can be extremely useful to you and your business. Learn more about how to utilize it with out Tableau in-person training today!