Machine Learning in SAP Analytics Cloud: Future of Decision Making Webcast Recap
Machine Learning in SAP Analytics Cloud – future of decision making – Webcast
Figure 1: Source: SAP
This was an SAP webcast from last month.
Direction going with the product
Key capabilities
New features
Figure 2: Source: SAP
Intelligent Era – evolution regarding IT systems; started with digitization of processes
Repetitive tasks can be automated; allows knowledge workers to work on high value tasks
More empowered employees
Figure 3: Source: SAP
Where SAP Analytics Cloud fits in the intelligent landscape
Embed machine learning in application
Open innovation platform to understand opportunities
Figure 4: Source: SAP
Challenges today
Data discovery failed to meet expectations
Analytics is subject to human bias; selection bias as you can’t ingest all information
How get to insights; typically through visual recognition; takes too much time
Next step to solve; plan, use machine learning – eliminates bias, uncovers insights
Figure 5: Source: SAP
Why not use machine learning? Too difficult, having to learn a program language, lack of expertise, lack of simple user interface/experience
In next 2 years, see 50% or more predictive done by citizen data scientist – not a job title, but an aspiration
Figure 6: Source: SAP
Make machine learning easy for business analysts
Product takes you on a journey – infuse results/techniques from machine learning, without having to take statistics courses
Start augmenting analytics with machine learning, get to data driven answers without human involvement – see key influencers, what is driving revenues, costs
Simulate change to take action
Start understand the patterns, trends, drivers in business
Feed trends into a business plan
Then assess
End to end cycle
Figure 7: Source: SAP
Digital Board room is the smart command center
Figure 8: Source: SAP
Content generation; in next 2 years, expect users to search for answers the way we naturally do
Search query in a natural sentence
Further enhanced with audio in the future
Also includes suggestions
Figure 9: Source: SAP
Where to get started
Classification is Y/N – will someone buy?
Regression – how explain something? What are variables contributing to revenue
Time series – here what has happened historically, project to future
Figure 10: Source: SAP
Think of how data needs to be manipulated
Figure 11: Source: SAP
Think of use case, where get data from and get data and source it
Subset of Q&A
Q: Are these available for live connections?
A: Not today; plan for end of year
Q: Does this have integration with different sources?
A: Depends on features
Smart predict – only for flat file only
More later this year
Q: When smart predict?
A: Net new customers – able to provision
Smart predict – new infrastructure – support for existing customers – working out what migration path looks like
Q: What are the machine learning capabilities?
A: Is it end user building models or computer building the models
Machine learning foundation – audio, image recognition – not in the platform
Depends on how you define
I'm always fascinated by this phrases we always hear when AI is part of conversation : "Analytics is subject to human bias; selection bias as you can’t ingest all information"
Why would "AI/Machine learning" have no bias ?
It is made by humans and operate based on set of algorithms that defy what it is supposed to do - the human bias is built in.
or is it ?
Is ultimate truth with no bias even possible ?
Hi Denis - that is a good point. One recent example was the 2016 election, where all the "polls" said it was 99% for one side and then the opposite happened.
I guess it depends on the algorithm and data too; where I work, machine learning says 9% of our commercials customers are more likely to go paperless...but I didn't inflect any bias.
Thank you for reading and commenting.
oh "paperless" - been hearing that for the last 20 years.....
still get incidents "we can't print our weekly reports" 😉