From Data Discovery to Adaptive Decision Making – Webcast Notes
This was a TDWI (Data Warehouse Institute)/SAP webcast with Dr. Barry Devlin of 9sight Consulting. The link to the slides is here.
Dr. Devlin started the webcast talking about Decision Making Support – where we came from and where we want to go to
What is the process behind them; we’ve tended to ignore
Process needs to be adaptive
Figure 1: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved.
Process has always been a part of BI
30 years later, now we have the Internet of Things
Privacy and security are important
There are new business models and new analytics
Now encourage better driving by linking better driving to cheaper insurance
New business models are based on data, based on action, small or large
Figure 2: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved.
Biz-tech eco-system is the view of where we are going with business where innovation is driven by technology; where the ability of IT and business can work synergistically together to use technology
A huge amount of change, externally sourced information, variety of information, mobile devices, customer interactions real-time, a lot of competition drives action
Data alone is no longer sufficient, data discovery is no longer sufficient
Figure 3: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved.
When you look at decision making, you see different processes – it – ad-hoc exploration – on the right an event happens, a drop in sales, problem with machine – what going to do? Starts by gathering information, innovation and use in action. Usually don’t have enough data or right information, then we iterate and go through the loop until we get action. Traditionally this has been done with spreadsheets. See the emergence of data discovery tools. The challenges include consistency and quality – issues in spreadsheet, right data, consistent and quality data. Then move to production – spreadsheet gets distributed by e-mails, moves to production in different forms. Now we have the “ Spreadmart challenge”
BI has not wanted to deal with using data warehouse
Figure 4: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved.
As Figure 4, shows an event comes in, conditioning and iteration is locked in IT box – ETL, data is high quality, fit for production
From the user, the only way to iterate is to go through IT
This is more like waterfall approach
BI tools became report generators, with slow iteration
Figure 5: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved.
Two models – one is good for innovation and the other production – bringing them together to make a decision support process
The way we run delivery support process
Process – these days of fast changing business, the expectation of analysts, data scientists, be innovative first by exploring information and data
Picture on right in Figure 5 shows the inner circle – “E” for exploring
Main emphasis is in the next two levels – cultivating – bring in peer collaboration, more structured way
Collaboration is done and shared, work together to improve, right data
Look at from point of view of quality control
The Grounding phase moves process to IT to apply centralized control
When queries / reports become popular, well-used, should in IT begin to see queries become important to marketing department; become a part of the ongoing process of department. Need to get ready for production – productionalize reports
Only start doing formal conditioning – particular set of reports, queries, become popular, need to take care that data is production-ready
Integration of data discovery and BI and collaboration with metadata; understanding the cycle is what the metadata cycle is
This is the model to use to look at the evolving tool landscape
Figure 6: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved
Figure 6 shows evolving tool landscape – data discovery tools adding data management and discovery features to “productionalize”
How bring them together
Figure 7: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved
Figure 7 shows the need to balance competing needs; first focus on decision making process
In new biz-tech eco-system cannot ignore decision making process
Sense and response MEDA model – see the book
- – How do we create a new way of being sensitive to what is going on in environment
Integrate collaborative work – social media and mining give ability to see what is going on
Mobile technology records the informal information which used to happen at the water cooler but part of innovative thinking – getting recorded in tweets, SMS messages, conversations on phone – becomes available for behavior mining. Scary as “big brother”
Predictive and other analytics, when have Hadoop, data scientist – Dr. Devlin believes predictive, data discovery, report generating – belongs in same area such as a BI Center of Competence
Figure 8: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved
Discussions can be in the middle of book
People, process, information
People make decisions, need to focus on how make them
Process – logic that is predefined and emergence – decision making is a process
Information – think of idea – difference between data and information where does knowledge come from
Take a look at the book in Figure 8.
Figure 9: Source: Slides reprinted by permission of TDWI (www.tdwi.org). Copyright (c) 2014 by 1105 Media. All rights reserved
Decision making is in a transition with biz-tech, new processes and doing business
Adaptive decision cycle – unite edge-on and center out models; make friends with spreadsheet world and data warehouse world – innovation and consistency unite
Process of decision making becomes key to business success – need to understand process people go through to make decisions
Tools are moving toward a unified tool
Q: How position data discovery in larger world of BI in general?
A: Depending on who you talk to – data discovery is a subset of BI
BI covers reporting, analytics, data discovery, preparation work that end users might to
Marriage of BI and Big Data – Webcast Summary
Why Big Data Analytics needs BI Too
Submit Early, Submit Often for ASUG Annual Conference 2015 – Top 10 Tips to Submit