Skip to Content
Author's profile photo Tammy Powlas

Burn that Churn? Checking Retention of Customers in Predictive Analysis

Do you want to see which types of customers are staying or leaving you?

The SAP Help says the R CNR Tree algorithm “Classifying and predicting one or more discrete variables based on other variables in the dataset”.  The help also states:

“Use this algorithm to classify observations into groups and predict one or more discrete variables based on other variables. However, you can also use this  algorithm to find trends in data.”

First, I have a set of data of customers by age, wealth, and savings.  I read that into SAP Predictive, and then select the R-CNR tree algorithm.

/wp-content/uploads/2012/12/1figconfigureclassificationregressiontree_164185.png

I’ll set the churn flag as the dependent column, the deposits, investments and independent, and use the classification method.

/wp-content/uploads/2012/12/2figsuccess_164222.png

The above shows I have run the analysis.

/wp-content/uploads/2012/12/3predval_164223.png

On the far right of the above, I can see the Predicted Values

/wp-content/uploads/2012/12/4stdcharts_164224.png

The above shows the standard charts and trends

5RUN.png

I run the algorithm again

/wp-content/uploads/2012/12/6predict_164226.png

/wp-content/uploads/2012/12/7count_164227.png

Looking at the data via Visual Intelligence, you can see which customers are you retaining and which ones are leaving.

/wp-content/uploads/2012/12/8youngandwealthychurn_164231.png

You can see that younger people are more apt to “churn” whereas individuals in older groups are more likely retained.

Assigned Tags

      4 Comments
      You must be Logged on to comment or reply to a post.
      Author's profile photo Víctor Abad García Sosa
      Víctor Abad García Sosa

      Excelent demo!

      Can you provide the data that you use to train the model?

      Author's profile photo Tammy Powlas
      Tammy Powlas
      Blog Post Author

      Hi - thank you for commenting.

      Sadly I only used this on a test laptop, which I had to return so I don't have that data anymore.

      Regards,

      Tammy

      Author's profile photo Former Member
      Former Member

      Tammy,

      This post was very useful in helping me to understand the use cases for this algorithm. The use case I have in mind involves finding trends, so I think I will be giving this one a try.

      Thanks,

      Gretchen

      Author's profile photo Tammy Powlas
      Tammy Powlas
      Blog Post Author

      Thank you, Gretchen - I did this a while ago.  I haven't downloaded another trial of Predictive since I am using Lumira for other things now, and both can't be installed on the same machine.

      Best of luck on your use case