Skip to content

Using Insights

Subscription Tier Required

This feature requires the Premier subscription tier or higher.

Instead of manually grouping, sorting, filtering, and drilling into data, GS Insights does this work for you.

Insights are automatically calculated whenever a Control Chart, Histogram, SPC Data Table, or Pareto Chart is loaded in a Dashboard.

Viewing Insights

When an Insight is detected for the first time, a notification will appear next to the Insights button. You will also be prompted to select how you would like to be notified of Insights in the future. This preference can be updated anytime from the User Settings overlay.

An image showing a chart with the insights button highlighted and a confirmation window asking how to alert the user in the future

You can click the Insights button at any time to open the Insights overlay. The types of Insights shown will depend on the chart’s data source (SPC or DMS).

SPC Insights

SPC Insights are automatically generated from Control Charts, Histograms, and SPC Data Tables. There are two types:

  • First-Level Insights automatically detect significant differences in variation or centerlines.
  • Second-Level Insights examine interactions between multiple Traceability fields.

First-Level Insights

In the guide on root cause analysis, you learned how to group control charts to discover causes of variation and process shifts. First-Level Insights replaces this manual work by automatically grouping the data by each Traceability field and searching for statistically significant differences. If differences are found, they are reported in the overlay.

An image showing the Insights overlay with a few insights

In the overlay, Insights are broken down into separate lists by Characteristic (or Short Run Standard), with the most impactful Traceability at the top.

  • If the Traceability is associated with the observed variability, the variation (spread) of the groups are significantly different.
  • If the Traceability is associated with the change in mean, the centerlines of the groups are significantly different.

Each row also has a Show Chart button. Clicking this button for a discrete Traceability draws a Control Chart grouped by that Traceability.

An image showing discrete first-level insights as grouped control chart

Clicking this button for a continuous Traceability draws a scatter plot of the data.

An image showing continuous first-level insights as scatter plot

Second-Level Insights

Sometimes, no single Traceability field reveals a root cause. Second-Level Insights looks for variation and process shifts related to the interaction between two Traceability fields. Access Second-Level Insights by clicking Explore Effects from the Insights overlay.

An image showing Explore Effects button

Select a Characteristic and a set of Traceability to analyze. GS will group the data by each combination of values from the selected fields and test whether those groups have significantly different means or variation. If both Traceability are discrete, the results are shown in a box-and-whisker plot. If one Traceability is discrete and the other is continuous, the results are shown as a scatter plot with the most impactful discrete Traceability highlighted.

An image showing results from the second-level insights overlay

Example

Suppose you are analyzing Hold Strength for molded plastic parts. Your data includes two Traceability fields:

  • Machine, with values Machine ABC and Machine XYZ.
  • Cavity, with values Cavity 1 and Cavity 2.

You have already reviewed the automatic First-Level Insights, but they did not reveal any significant differences between different cavities or different machines.

But after selecting Hold Strength in the Second-Level Insights overlay and clicking Analyze, GS will group the data into four groups to analyze:

  • Machine ABC and Cavity 1
  • Machine ABC and Cavity 2
  • Machine XYZ and Cavity 1
  • Machine XYZ and Cavity 2

After processing, GS reports that Combinations of Cavity and Machine are significantly different. Clicking the Show Chart button pulls up the following box-and-whisker plot with some clear differences:

  • Two green boxes, showing a decrease in variation in these groups.
  • A dark red box, showing an increase in variation and a significant shift in mean for the Machine XYZ, Cavity 2 group.
  • A light red box, showing an increase in variation for the Machine ABC, Cavity 1 group.

An image showing box-and-whisker chart for combinations of cavity and machine

Box-And-Whisker Plot Colors

Colors in box-and-whisker plots are used to represent changes in variation and mean.

Color Change in Variation Change in Mean
Green Decreased variation No significant change
Yellow No significant change Significant change
Light Red Increased variation No significant change
Dark Red Increased variation Significant change
Gray No significant change No significant change

DMS Insights

In the guide on root cause analysis, you learned how to group and drill down in Pareto charts to discover probable causes of Defects. DMS Insights replaces this manual work by automatically grouping the data by every Traceability and searching for groups with a significantly high DPM (Defects per million) or Cost per Unit. If any significant groups are found, they are reported in the overlay. Insights are broken down into separate lists by each graph in the Pareto chart, with Percent Defects and Cost tables.

An image showing the DMS Insights overlay with a few insights

Each row includes a Show Chart button. Clicking this button for a discrete Traceability draws a Pareto chart grouped by that Traceability.

An image showing discrete first-level DMS insights as pareto chart

Clicking this button for a continuous Traceability draws a scatter plot of the data with the Traceability value on the x-axis and the DPM (or Cost per Unit) on the y-axis.

An image showing continuous first-level DMS insights as scatter plot

See Also