ISACA Now Blog

Knowledge & Insights > ISACA Now > Posts > Understanding Big Data and Machine Learning Projects

Understanding Big Data and Machine Learning Projects

Folakemi Gbadamosi, MCS, CGEIT, CRISC, COBIT 5 F-I-A
| Posted at 2:59 PM by ISACA News | Category: COBIT-Governance of Enterprise IT | Permalink | Email this Post | Comments (3)

Folakemi GbadamosiBig data and machine learning have rocketed to the top of the corporate agenda. Executives look with admiration at how Google, Amazon and others have eclipsed competitors with powerful new business models derived from an ability to exploit data. They also see that big data is attracting serious investment from technology leaders such as IBM and Hewlett-Packard. Meanwhile, the tide of private-equipment and venture-capital investments in big data continues to swell.

AI/machine learning also continued to rise toward the top of technologies considered to have the highest potential to deliver transformative value to organizations. While placing second in these rankings according to ISACA’s 2018 Digital Transformation Barometer, AI/machine learning went from 18 points behind big data in 2017 to just 3 points behind big data in 2018. As the perceived value of AI continues to increase, the proportion of organizations planning to deploy AI continues to increase as well, with a 35 percent increase over the 2017 report.

What audit, risk, assurance and security practitioners and executives should know about big data and machine learning projects
Perhaps you have heard about a new algorithm that can drive a car? Invent a recipe? Detect fraud ? Scan a picture and find your face in a crowd? It appears every week that companies are discovering new uses for algorithms that adapt as they encounter new data. Machine learning has tremendous potential to transform companies, but in practice it is usually far more mundane than robot drivers and chefs. Think of it simply as a branch of statistics, designed for a world of big data. Executives who want to get the most out of their companies’ data should understand:

  • What it is
  • What it can do
  • What to watch out for when using it

The enormous scale of data available to firms can pose several challenges. Of course, big data may require advanced software and hardware to handle and store it. Machine learning is about how the analysis of the data must also adapt to the size of the dataset. This is because big data is not just long but wide as well.

Big data projects versus traditional IT projects
“90% of the effort in successful machine learning is not about the algorithm or the model or the learning. It’s about the logistics.”
From Machine Learning Logistics by Dunning and Friedman (O’Reilly, 2017)

Logistics are not the only issue that matters for success. Connecting AI and machine learning projects to real business value is of huge importance. The social and cultural structures of your organization make a big difference, as well.

The following table shows the distinction between big data and traditional IT projects, tapping into COBIT 5 components (Five Principles, Seven Enablers, Trigger Events, Pain Points and the seven phases of Program Management used in the life cycle model).

BIG DATA PROJECT

TRADITIONAL IT PROJECT

TYPICAL PAIN POINT/TRIGGER EVENT

Develop a new shared understanding of customers’ needs and behaviors

Predict future growth markets

Install an ERP system

Automate a claims-handling process

Optimize supply chain performance

ENABLER 4 CULTURE, ETHICS AND BEHAVIOR

Change how employee think about use of data

Challenge the assumptions and biases employees bring to decision-making

Use new insights to serve customers better, build new businesses and predict outcomes

Improve efficiency

Lower costs

Increase productivity

THE SEVEN PHASES OF PROGRAM MANAGEMENT USED IN THE LIFE CYCLE MODEL

DISCOVERY-DRIVEN PROJECT MANAGEMENT:
Develop theories

Build hypotheses

Identify relevant data

Conduct experiments

Refine hypotheses in response to findings

Repeat the process

TRADITIONAL PROJECT MANAGEMENT:
Define desired outcomes

Redesign work processes

Specify technology needs

Develop detailed plans to deploy IT

Manage organizational change and train users

Implement plans

ENABLER 7 – PEOPLE, SKILLS AND COMPETENCIES

IT professionals with engineering, computer science, and math backgrounds (in some cases)

People who know the business

Data scientists

Cognitive and behavioral scientists

IT professional with engineering, computer science, and math backgrounds

People who know the business

CHALLENGES TO SUCCESS: DID WE GET THERE AND HOW DO WE KEEP THE MOMENTUM GOING?

Employee bases decision on data and evidence

Employee uses data to generate new insights in new contexts

Project comes in on time to plan, and within budget

Project achieves the desired process change


In conclusion, big data and machine learning projects involve new technology and new development approaches, and are inherently risky. If you are doing significant data exploration or discovery with big data, you will occasionally fail—which is not really a problem if you learn from the failures. Big data and machine learning projects are still more like R&D than production applications.

Comments

Re: Understanding Big Data and Machine Learning Projects

This is an insightful post. Thank you
Francis_O at 11/2/2018 9:44 AM

Great stuff

What a great, thought-provoking article on AI. With AI requiring new approaches and thinking, will boards be ready to embrace unpredictability of the output?

Not all R&D output can be immediately commercialised, will boards be willing to accept this?
Joe_Kutoane at 11/5/2018 2:55 AM

Great stuff

What a great, thought-provoking article on AI. With AI requiring new approaches and thinking, will boards be ready to embrace unpredictability of the output?

Not all R&D output can be immediately commercialised, will boards be willing to accept this?
Joe_Kutoane at 11/5/2018 2:55 AM
You must be logged in and a member to post a comment to this blog.
Email