Skip to Content

Predictive analysis capabilities have been available in Lumira since release 1.25. This easy to use feature can add significant value to your analyses.

As an example, the chart below shows monthly gas consumption over a 3.5 year period. Not unsurprisingly, consumption is low in Summer and high in Winter.

/wp-content/uploads/2015/12/predict1_849214.png

We will use predictive analysis to predict gas consumption for the next 6 months, based on the existing historical data.

Click on the option button for the appropriate measure (in this case it is Consumption) and choose “Predictive Calculation”. Two types of calculation are currently available – forecasting and linear regression. We will look at forecasting first.



Forecast

The Forecast dialog offers two forecasting algorithms – the formula from SAP Predictive Analytics and Triple Exponential Smoothing. The former is preferred, but it requires a minimum number of data points. The latter will always give a result but is less accurate and should only be used in scenarios where there are not enough data points for the SAP Predictive Analytics forecast algorithm.

/wp-content/uploads/2015/12/predict2_849236.png

The blue line is the calculated forecast. We notice that it tracks the historical data (green line) quite accurately so this gives us confidence in the 6 months forecast data.


Linear Regression

Next we will add a linear regression calculation. A linear regression identifies the underlying linear trend in non-linear data.


Click on the option button for the appropriate measure (in this case it is Consumption) and choose “Predictive Calculation”. This time we will select Linear Regression.


In the Linear Regression dialog there is just one option, the number of time periods to extend the prediction. For this gas consumption example I will extend the prediction for 6 months.


/wp-content/uploads/2015/12/predict3_849237.png


The orange line is the calculated linear regression. In our example we can see that, over time, gas consumption is slowly decreasing.



Further analysis


It would be interesting to explore what factors influence the decreasing gas consumption. One option would be to blend daily temperature data with gas consumption. For hints and tips on data blending, please refer to Did you know?… How to do Data Blending in Lumira.

“Did you know…” is a series of short blogs by the BI & Analytics Competence Centre, a global team within SAP engineering focused 100% on customer success. They are a useful reminder of Lumira hints, tips and best practices.

To report this post you need to login first.

Be the first to leave a comment

You must be Logged on to comment or reply to a post.

Leave a Reply