Skip to Content
Product Information

Feature Highlight: Smart Insights and Smart Discovery Session Recap

This was an SAP webcast from last week. I won’t repeat the slides from previous blogs. The webcast covered the features of Smart Discovery and Smart Insights.

Don’t forget to join us for these upcoming related webcasts:

February 13 BI: What’s New: Data Wrangling and Modeling in SAP Analytics Cloud

February 14 BI: Empowering Business Users with Analytics On The Go

February 19 BI: SAP Analytics Cloud for BW Customers

February 26 BI: Enabling SAP Analytics Cloud for SAP BusinessObjects BI customers

Smart Discovery

Smart Discovery helps “how we find out why”, themes in data and make better decisions

Augmented analytics is the analytics of future -take what do, more productivity, get decisions faster

Automate processes, using ML and AI, creativity, to move business forward

1smart2019.jpg

Source: SAP

Is this useful? See factors influencing information

See interesting patterns, anomalies

This is a new iteration of Smart Discovery; automatically creates story

2smart19.jpg

Source: SAP

4 pages of information

Overview gives broad information of data

Key influencer page tells you what other variables influence data

Unexpected values – data that do not follow the pattern; show anomalies

Simulation – see what would happen if you add a new row of data

3smart2019.jpg

Source: SAP

How it works

If select a measure, a regression model is built

If a dimension is selected, classification model

4smart2019.jpg

Source: SAP

Builds model; SAP has a unique automated machine learning technology, successfully tuning models

Can build model without input of a data scientist

Allows business users to create predictive models to improve decision making

The slides below were covered in the demo, so no notes:

5smart2019.jpg

Source: SAP

Source: SAP

8smart2019.jpg

Source: SAP

9smart2019.jpg

Source: SAP

10smart2019.jpg

Source: SAP

Smart Insights

Smart Discovery selects a variable, notices a pattern

Smart Insights – pick a data point, variance in data, and understand what is behind that data

11smart2019.jpg

Source: SAP

Smart insights helps find quickly what is behind a particular item

It can add contexts to your visualization to help understand what is going on

12smart2019.jpg

Source: SAP

“Understand why, faster”

Automatically describe drivers behind a data point

Sometimes it will be obvious

13smart2019.jpg

Source: SAP

You can see what influences a data point, the drivers

You can do the same with a variance and run smart insights

You can embed smart insights in a chart

14smart2019.jpg

Source: SAP

What Smart Insights tells you when you select a data point

Q&A

Q: Can we export the smart insights to an Excel or PDF?

A: Not at the moment

Q: Can you compare on-premise Predictive with Predictive in SAC?

A: Predictive Analytics (on-premise) was a tool for data scientists; the user is a data scientist

Smart features in SAC are aimed at the SAP Analytics Cloud user, the business user, the business analyst

Underlying technology in automated on-premise and SAC are the same but targeted personas are different

Q: Supported in other languages?

A: Yes, works in 16 supported SAC languages

Q: Are these functions available for live connections?

A: Available for imported data, working on live connections

Search to Insight supports Live HANA connection

Q: Will saving simulations be on the roadmap?

A: not at the moment, this is an exploratory tool; look at the Smart Predict

Q: What kind of license is required?

A: Smart features are part of the standard BI license

Q: What is insight quality based on?

A: Calculated on how well does the data explain target value, and other is if supply other data, how well is it distributed, how performs against random and perfect model

Q: Can you select multiple measures for Smart Discovery?

A: No, looking at adding more

Q: Smart discovery data limitation?

A: Supports up to 1 million cells

 

What do you think?

3 Comments
You must be Logged on to comment or reply to a post.
  • Dear Tammy,

    I just read your blog and ran the Smart Discovery on a new data model I created last week.

    Firstly, I would like to make a suggestion.  We need to be able to add tooltip dimensions to the charts that Smart Discovery generates.  In some graph types such as the scatter-plot this option is not available and the the tooltips generated leave out an important dimension that my clients will surely ask to see.

    Also some content can still not be copied to other pages in the story, which is disappointing.

    The other point I would make is that I think we are still quite a long way from the results of Smart Discovery being completely easy to understand for the end user and a lot depends on selecting a truly representative selection of dimensions in order to not produce confusing results for certain graphs.  The waterfall chart in the Simulation section is particularly prone to this.  All the comparison graphs that use data bins in the Key Influencers section are not that intuitive to follow either.  I have had issues explaining this to end users that do not have a statistical training.  This opacity creates a responsibility on the part of implementing partners to provide training and support to end users.  In turn it would be great if SAP could arrange specialised courses for implementing partners so that we are best prepared for this task.  Some simple case study data would be great as well as I have clients that want to be able to check the calculations on everything so they can feel sure about the results.

    On the plus side I am very impressed with the Unexpected Values section and I will try and apply the findings to forthcoming projects to demonstrate its usefulness.  I think the Simulation has great potential too but feel at present that I need to run a lot of experiments on it to understand it better.  My earlier point about correctly selecting the measures and dimensions seems important here as what increases and decreases the expected results is by no means convincing in the data model I am using.

    Thanks for the blog post!

    Kind regards

    James Longstaff

     

    • Hi James – good feedback, thank you for reading and replying.  I recommend that you suggest improvements to Smart Discovery at the SAP Influence site at https://influence.sap.com/sap/ino/#/campaign/884