This transcript comes from the online course, Machine Learning Leadership and Practice – End-to-End Mastery.
In 57 words, here’s why machine learning’s important: Business needs prediction. Prediction requires machine learning. And machine learning depends on data. Putting that in reverse, we have data, we give it to machine learning, it makes models that predict, and we use the predictions to improve all the main stuff we do, all the large-scale operations of organizations that make the world go ’round.
data → machine learning → model → predictions → operations
In this article, I’ll run through these five concepts to show you how learning from data to predict revolutionizes business and I’ll make the case that machine learning is the coolest and most interesting science, period. And then, in the next piece, I’ll reveal why this specialization of three courses is a unique, effective way to get started with machine learning.
Ok, here’s the opportunity: The world is a remarkably inefficient, wasteful place. The organizations that treat and serve us as consumers constantly get it wrong. Most mail is junk mail. Institutions are blindsided by risky debtors and policyholders. Lot’s of fraud goes undetected and yet most of the transactions humans bother to audit are legit and don’t actually need auditing. In healthcare research, we typically show that a treatment works in general, but we can’t discern which patients would actually be better off untreated.
These are heavy costs that tax both you and I every day. If only there were some way to run things better, to do a better job making all these decisions, to improve the effectiveness of the frontline operations that define a functional society. Well, I just happen to have a suggestion.
Prediction as a capability is the Holy Grail for driving decisions such as whom to target for marketing, for investigation, or for protection from financial or medical risk. Now, perfect prediction is not possible, but even lousy predictions that are at least better than guessing often deliver a tremendous systematic benefit. Playing these numbers games better, tipping the odds even just a bit more in our favor, generates an enormous impact.
And this is where the idea of learning comes in. It learns to predict. Kinda like what people do — I’m drawing an analogy — an organization learns from its overall cumulative experience when its computer learns from its data. The data’s a record of things that have happened, the aggregate experience. Advanced number crunching methods that discover patterns in the data are what we call machine learning.
That learning process makes it possible to predict — to put odds on behaviors and outcomes, such as whether a customer will click, buy, lie, or die. In fact, every important thing a person does can be valuable to predict, including: consume, work, quit, vote, love, procreate, divorce, default on credit payments, cheat, steal, or kill.
The predictions from machine learning drive millions of decisions more effectively, determining whom to call, mail, approve, test, diagnose, warn, investigate, incarcerate, set up on a date, or medicate. Improving these many decisions even just a bit delivers a huge win.
And prediction’s not the only capability. Computers also learn to identify objects in images — which is key for self-driving cars, smart manufacturing, medical image processing, and even Facebook’s suggestions for tagging photos — and they learn to identify words within sound for speech recognition. These capabilities have made leaps and bounds in recent years, mostly due to a specific machine learning method called deep learning.
So, by the way, machine learning lives at the center of “artificial intelligence,” and so this is very much an AI course. But, AI really is just a nifty story, not a well-defined technology. Often, when people say AI, they actually just mean machine learning.
Anyway, the outlook for machine learning is better than ever because… it’s raining data. Every credit card transaction, Facebook post, medical diagnosis, car accident, and sales call is recorded. Data grows by an estimated 2.5 quintillion bytes per day. It’s the most potent “unnatural resource.” Today’s totally historical advent of having data about everything and using data for everything is a profound game changer.
Now, data does seem boring to some people. If so, I’d argue, you’re not geeky enough. I’ll change your mind by showing you that data isn’t arcane — after all, it’s actually a long list of things that have happened — and I’ll show you that the discoveries from data make actual sense, like sometimes even in the form of if-then business rules.
And besides, it’s freakin’ cool! When you unleash the power of data in this way, you’re making the ultimate use of these amazing general-purpose machines that we know as computers, using them for the most all-encompassing of tasks, which is: getting better at tasks — that is, learning. The algorithms to do this take on the most transcendental kind of scientific challenge: to generalize from examples and discern truths that hold in this world.
Beyond a field of science, machine learning is a movement that’s exerting a forceful impact. It reinvents industries and runs the world. It’s come of age as a pervasive business practice necessary to thrive and even just survive. Companies, governments, law enforcement, and hospitals seize upon this power to boost sales, cut costs, combat risk, prevent fraud, fortify healthcare, streamline manufacturing, conquer spam, toughen crime fighting, and win elections.
Your team needs it, your boss demands it, and your career loves machine learning. After all, it’s consistently ranked as a top in-demand skill and LinkedIn places it as the very top emerging job in the U.S.
Coursera co-founder Andrew Ng calls machine learning “the new electricity,” and The Harvard Business Review calls it “The most important general-purpose technology of our era.”
I would like to hereby designate machine learning the most fascinating, promising, and exciting branch of science and technology, period. It’s the coolest.
About the Author
Eric Siegel, Ph.D., is a leading consultant and former Columbia University professor who makes machine learning understandable and captivating. He is the founder of the long-running Predictive Analytics World and the Deep Learning World conference series, which have served more than 17,000 attendees since 2009, the instructor of the end-to-end, business-oriented course Machine Learning Leadership and Practice – End-to-End Mastery, a popular speaker who’s been commissioned for more than 100 keynote addresses, and executive editor of The Machine Learning Times. He authored the bestselling Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die, which has been used in courses at more than 35 universities, and he won teaching awards when he was a professor at Columbia University, where he sang educational songs to his students. Eric also publishes op-eds on analytics and social justice.
You must be logged in to post a comment.
Pingback: Why It’s Vital and How to Do It « Machine Learning Times – Machine Learning
Pingback: Real-Time Machine Learning: Why It's Vital and How to Do It - Predictive Analytics World