As digital streams drained into oceans of data, a new age was born — the Digital Age. It is no coincidence that machine learning matured rapidly in this new age. Ishan Gupta (@ishan_gupta), Managing Director of Udacity India, asserts machine learning is no longer a buzzword but a permanent feature of today’s business world. “Machine Learning,” he writes, “will come to change the way you work, earn a livelihood, purchase and consume goods and services. … What one needs to realize is that what the personal computer and the smartphone were to Steve Jobs, machine learning (ML) could be for you — but you need to start now.”[1] Although that call to action sounds a bit like embracing machine learning just because it’s new, Gupta knows well a business case must be made for the implementation of any new technology. That shouldn’t be hard. Janakiram MSV (@janakiramm), Principal Analyst at Janakiram & Associates, explains, “ML is becoming the front and center of many emerging technologies including Cognitive Computing, Artificial Intelligence, Chatbots, Personal Assistants, and Predictive Maintenance.”[2]

What is machine learning?

MSV asserts machine learning is all about math, statistics, data, and programming. Programming, however, in the form of algorithms, is where the rubber meets the road. Jen Underwood, founder of Impact Analytix, observes, “Anyone can be empowered to rapidly extract value from data using a plethora of algorithms to make better decisions, act quickly and achieve better outcomes.”[3] She warns, however, “Machine learning is susceptible to a wide variety of bias types and a myriad of other issues if applied improperly. It also has amazing untapped potential when implemented correctly.” Analysts at McKinsey & Company explain, “Machine-learning algorithms detect patterns and learn how to make predictions and recommendations by processing data and experiences, rather than by receiving explicit programming instruction. The algorithms also adapt in response to new data and experiences to improve efficacy over time.”[4]

What’s really valuable about machine learning — and why it’s here to stay — is that it can provide predictions and prescriptions. McKinsey analysts explain there are three types of analysis — descriptive, predictive, and prescriptive — with each type of analysis being more complex than the one before it. The most common type of analysis is descriptive. McKinsey analysts explain descriptive analysis can describe what has happened in the past. They also note this type of analysis is widely used across industries. Predictive analytics ramp up complexity and can anticipate what will happen in the future. They note predictive analytics are used by data-driven organizations as a key source of insight. Ramping up complexity once more, machine learning can perform prescriptive analytics. McKinsey analysts explain prescriptive analytics provide recommendations about what to do to achieve goals. They note this type of analysis is used heavily by leading data and Internet companies.

McKinsey analysts go on to explain three major types of machine learning: supervised learning; unsupervised learning; and reinforcement learning. Cynthia Harvey defines those types of machine learning this way:[5]

• Supervised Learning: “Supervised learning requires a programmer or teacher who offers of examples of which inputs line up with which outputs.”
• Unsupervised Learning: “Unsupervised learning requires the system to develop its own conclusions from a given data set.”
• Reinforcement Learning: “Reinforcement learning involves a system receiving feedback analogous to punishments and rewards.”

“For business and data analysts,” Underwood writes, “most augmented and automated solutions leverage well-known, supervised, unsupervised and deep learning algorithm libraries behind the easy buttons. They merely simplify the machine learning life-cycle process in business user-friendly apps. The differences in algorithm performance can usually be evaluated and measured by precision, accuracy and other criteria.”

Machine learning is only as good as the data received

Data is the lifeblood of the Digital Age. Data, however, is not all of the same quality. From the beginning of the computer era, computer scientists have droned the mantra, “Garbage in, garbage out (GIGO).” According to Wikipedia, “The first use of the term has been dated to a November 10, 1957, syndicated newspaper article about US Army mathematicians and their work with early computers, in which an Army Specialist named William D. Mellin explained that computers cannot think for themselves, and that ‘sloppily programmed’ inputs inevitably lead to incorrect outputs.” GIGO is especially true as it relates to machine learning. Thomas C. Redman (@thedatadoc1), President of Data Quality Solutions, explains, “Poor data quality is enemy number one to the widespread, profitable use of machine learning. While the caustic observation, ‘garbage-in, garbage-out’ has plagued analytics and decision-making for generations, it carries a special warning for machine learning. The quality demands of machine learning are steep, and bad data can rear its ugly head twice — first in the historical data used to train the predictive model and second in the new data used by that model to make future decisions.”[6]

When dealing with clients, I often have to stress the importance of data in achieving the results they desire. If machine learning solutions don’t have the right data, they simply can’t provide the right answers. Redman explains, “To properly train a predictive model, historical data must meet exceptionally broad and high quality standards. First, the data must be right: It must be correct, properly labeled, de-deduped, and so forth. But you must also have the right data — lots of unbiased data, over the entire range of inputs for which one aims to develop the predictive model. Most data quality work focuses on one criterion or the other, but for machine learning, you must work on both simultaneously.” Underwood adds, “If an analyst or business user feeds an automated machine learning solution poor quality data, the predictive results will be poor. … There still is an art to designing and providing data that accurately reflects a business process even if automated analytics can work through millions of variable combinations that would be unreasonable for a human to do. Only a human can understand and decipher nuances in business context.”

Summary

Gupta believes machine learning will spark a revolution in the business world. “To affect this revolution,” he writes, “businesses across industries and geographical regions will need to update and reform their processes to integrate ML. Those that don’t adapt will fall behind, to be replaced and supplanted by newer and more dynamic companies that will use ML to drive their growth. This brave new world will be accessible to those who get on the ML bandwagon early.” Redman agrees machine learning can change the business landscape. “Machine learning has incredible power and you need to learn to tap that power,” he writes. He cautions, however, “Poor data quality can cause that power to be delayed, denied, or misused.” Machine learning is here to stay. Use it wisely.

Footnotes
[1] Ishan Gupta, “Machine learning isn’t just a buzzword anymore – it’s here to stay,” Your Story, 19 March 2018.
[2] Janakiram MSV, “Machine Learning Is Not Magic: It’s All About Math, Stats, Data, and Programming,” The New Stack, 7 April 2017.
[3] Jen Underwood, “My Algorithm is Better than Yours,” InformationWeek, 28 March 2018.
[4] Staff, “An Executive’s Guide to AI,” McKinsey & Company.
[5] Cynthia Harvey, “What is Machine Learning?” Datamation, 3 January 2018.
[6] Thomas C. Redman, “If Your Data Is Bad, Your Machine Learning Tools Are Useless,” Harvard Business Review, 2 April 2018.