Comparing Alert Data to Non-Alert Data in SAS Field Quality Analytics

[#1]

The purpose of this post is to show how to compare alert data to non-alert data in SAS Field Quality Analytics. As a reminder SAS Field Quality Analytics generates alerts when it identifies higher than usual events. A natural progression after an alert is generated is to compare data to where the issue is known to data where the issue is not present. This allows for the root cause to be identified quicker, leading to possible reduction in warranty-related costs.

In the alert generated below we know that there is an issue when customer country is the United States, and when the Primary Labor Group is I.

Select any image to see a larger version.
Mobile users: To view the images, select the "Full" version at the bottom of the page.

To identify the root cause of this alert it can be helpful to compare it to data where the alert is not present. To do so select Analyze alert.

Next, select the desired project and give a unique Data selection name. After the desired information is entered select Save.

The specified analysis will be created underneath the new data selection in the specified project. Open the created analysis.

First, select the cells around and including the alert. Second, select Analyze further and then Perform combined analysis.

Select the desired analysis type. In this example I will create a Pareto. Then select Save.

In the Pareto definition window change the Reporting variable to Alert, and the Group variable to Model Code. The Alert variable is a binary generated variable of Yes or No indicating, whether an alert is present. This variable could also be used in a different type of analysis such as a decision tree or a reliability analysis. Select Submit.

Once the Status is changed to Completed, open the Pareto to view the results. This analysis grouped Total Claim Cost by the three Model Codes Abyss, Galacto and Gemini. It also compares if there is an Alert to no Alert. The Orange in the graph represents when there is an alert. We can see in this output that the Total Claim Cost is highest when there is an Alert and the Model Code is Gemini. This helps us narrow our search to Gemini Models.

If desired, I could then analyze this data further, or I could analyze the existing subset on a different variable. For example, I could dig deeper into the Labor Code or the location of the events. This post focuses on using the Alert variable to compare Alert data to Non-alert data. Additional information can be found in the SAS® Field Quality Analytics on SAS® Viya® course which is part of the SAS Internet of Things Learning Subscription.

Find more articles from SAS Global Enablement and Learning here.