Introducing Key Driver Analysis - SaaS in 60

[Michael_Tarallo]

With key drivers, users can better understand the reasons for historical results, not just the what but the why, providing deeper insight and supporting more meaningful action. 

[Michael_Tarallo]

This week we introduce Key Driver Analysis,  a new capability that leverages the power of Qlik AutoML to surface the most influential factors that are driving an organizations outcomes in their analytics. Users can simply right click on a chart to initiate a key driver analysis and then specify a target measure and specific features they want to include.  The AutoML engine will then generate an analysis that shows the fields in your data model that are most affecting the target measure, complete with visualizations and Natural language generated insights.   You can see the key drivers ranked by importance, and can further select each one to get a breakdown and distribution of the values and their influence.  With key drivers, users can better understand the reasons for historical results, not just the what but the why, providing deeper insight and supporting more meaningful action

Resources:

• Qlik Cloud - What's New
• Key Drivers - Qlik Help

[CJ_Bauder]

@Michael_Tarallo  This is incredible! No doubt this is a great way to begin getting involved in the AutoML side of Qlik SaaS and propagate further conversations targeted at using AML to solve business issues in a realistic and meaningful way. Looking forward to digging further into the details.

Thanks!

[jfitz_chicago]

Great driver of business value!

[Anil_Babu_Samineni]

It was really nice to have such a huge data consumer. Thank @Michael_Tarallo for sharing a video.

[arychener]

Hi,

Thanks for this video!

Each new key factor analysis creates a new AutoML model, right? Does the number of models count in the authorized number of deployments (depending on subscription)?

Thanks a lot.

Amélie

[paulcalvet]

Hello,

I have a question 

In the documentation there is :

Distribution chart

The Distribution chart also breaks the feature down by unique value. Each instance of the feature value within a record in the dataset is visualized as a bubble. The bubbles are distributed and ranked according to influence on the target. You can toggle between average and total influence.

The influence values are shown as absolute values. Therefore, a feature value could have a strong negative impact on Sales, but you might see that it is shown as having the highest influence on this target.

--> But when I run a model, i can see negative value in the distribution chart, what does this value mean ?

Thanks,