Insight Analysis
This page describes the Insights Analysis capability.
Last updated
Was this helpful?
This page describes the Insights Analysis capability.
Last updated
Was this helpful?
The Automated Insights list is located on the Right Sidebar of the Advisor Workspace, and allows you to explore what happened in the data related to a specific Topic.
This information adds valuable insights, making it easier to understand answers to questions.
But this is not all: Crystal is able to dig even deeper!
You can request an Insight Analysis and get straight to the point of why and how the events described in the Automated Insight happened.
This in-depth exploration will help you understand your data better and make data-driven decisions without losing sight of any piece of information.
In particular, Insights Analysis is based on a Data Analysis technique called Diagnostic Analytics.
Pills of Data Analysis
Diagnostic Analytics are information that describe the reasons why an observed pattern or trend happened.
For example, they can describe what went well, what went wrong, and what had the most significant influence.
This is how you can request and receive an Insight Analysis:
After asking a question, find the list of Automated Insights on the right-hand side.
You can see that some Insights include an "Analyze" option.
Remember
Right now, Insight Analysis can be requested for all Topics but it's only available for the Top Contributor, Average and Variability and Extreme Value Insights.
click on the “Analyze” icon for the Insight you're interested in.
You will see a banner notifying you that the analysis has started, meaning Crystal is processing your analysis to uncover reasons behind the Insights.
This process may take several minutes, depending on the amount of data that needs to be processed.
You'll receive a notification once your analysis is complete.
Once you open the notification, you can access the result of the analysis in the intro message of the Topic.
Once you pick an Insight, Crystal identifies the appropriate Data Model for analysis.
A Data Model is a consolidated view of data tailored for specific analytics. It is useful to find relationships between data.
In this case, the Data Model is constructed by linking the various data inputs that led to the construction of the Topic. Crystal creates an extended version made by combining various data inputs used to create the original Topic.
Diagnostic Analytics will then use this extended Model, excluding the columns not used in all the Topic Configurations within the specific project.
This approach aims to highlight the main Factors behind an Insight, helping you understand the data better.
Right now, Insight Analysis can be requested for all Topics but it's only available for the Top Contributor, Average and Variability and Extreme Values Insights.
Insight Analysis is only available on Insights generated from Configured Topics, i.e. the ones that have been manually created by the Admin, whereas auto-generated Topics are not supported.
To access the Diagnostic Analytics, you first need to access the Automated Insights
The process of requesting and receiving an Analysis is asynchronous: the average waiting time is correlated to the DB size.
For a small or medium-sized DB, you can expect a processing time of around 5 minutes.
You can request only one Analysis at a time for each Card that supports the Diagnostic Analytics.
If you attempt to request an analysis that is already in progress, you will be informed that the analysis is still being processed and that you should wait for the dedicated notification.
You can't receive the analysis if the Topic from which the Insight was generated doesn't have a real Extended Data Model associated with it. This means that the original Joined Fact Table doesn’t contain any extra Entities other than the ones that are already available in the Topic.
A Diagnostic Analysis can’t be performed if a Topic only has categorical columns, as could be the case for Lists and Descriptions. However, applicability may vary depending on the Topic's specific configuration.
A Diagnostic Analysis can’t be performed if a Topic has at least one Custom Column.
If you don't have the permissions for certain underlying data, you will receive a partial analysis with that data hidden. The analysis will reveal information about the Dynamic Filter Value that drove the Insight, but inform you that more details aren't available due to missing permissions.
There may be a constraint regarding the number of rows the Feature can handle, though it is estimated to be around hundreds of thousands, chosen from the most recent rows in the considered Data Model of the analyzed insight.
The Diagnostic Analytics provide different possible outcomes, common to the Insights in focus.
In particular, Crystal can tell you:
1
Which Entity has contributed the most in achieving the result mentioned in the Insight (e.g. "Country").
2
Within the selected Entity, it can identify the one specific Entity Value that has greatest impact on the result mentioned in the Insight (e.g. "Italy").
3
How and by how much the result mentioned in the Insight would be impacted if the Entity Value with greatest impact were to be excluded.
Keep analyzing your data!
