In 2002, I saw an amazing movie, Minority Report with Tom Cruise. It really made an impression on me—not for the science fiction nature of it, but for the possibilities or the reality of it.
I work in machine learning, data mining, and applied math area. I work with a lot of very smart people here at Oracle, we do amazing things. And when I got out of that movie, I thought, "Hmm, that's pretty possible. I don't see that as being so farfetched."
The reason that the movie rings so true is Oracle's strategy, because at Oracle, our strategy is to move the algorithm, not the data. And when we do that, things like Minority Report, things like the type of new scenarios and possibilities that are written in Patrick Tucker's book, The Naked Future, all of those things become so much more possible. What happens in a world that anticipates your every move?
At Oracle, we've taken a different approach all together. We've said data gets bigger and bigger and bigger each year, and data, at some point, becomes so large that it becomes almost immovable, and it makes no sense to move all the data to some other location to calculate a median, or to do a T test, or to run a decision tree, or a logistic regression, or a neural network, or you name it, whatever. What makes much more sense is to bring the algorithms to the data, and that's what we've done.
So, imagine today, you're on your iPhone, and you wake up in the morning. I'm in the Boston area, and I get the local news, and I get some latest update about something that's going on in Boston. And I might get the local news, the local weather, might be my stated interests, my favorite sports teams, the Boston Celtics, the New England Patriots, the Red Sox, that kind of stuff, and I might get some national news updates. Traditional stuff, right?
Now, imagine a slightly revised future based on an example in The Naked Future. I wake up and my device tells me, "When you meet your old girlfriend at the coffee shop this morning, act surprised to learn that she's getting married."
Huh, that's interesting.
So, I meet my old girlfriend at the coffee shop and I say, "Oh, by the way, congratulations on getting married," and she sort of recoils and says, "What do you mean I'm getting married? Who told you that? How did you know?" And he goes, scrambling, I say, "Well, I don't know, I think I saw it on your Facebook post." And she says, "I didn't post anything to anybody anywhere. How did you know? How did you know?" And it becomes kind of confrontational.
And so, if you look at the very near future, the possibilities there, you can see how this could have easily happened. You're collecting a lot of different data from different locations, different places, and you have maybe the girlfriend changed her address recently, maybe she's moved in with a boyfriend or moved out of the house into an apartment, or who knows what. Maybe they've recently adopted a dog, maybe they've had a lot of Facebook pictures of the two of them together, maybe there's some tweets of, "I'm so in love," things like that, "Looking forward to spending our life together forever," things like that. And maybe there's an online ring purchase off Amazon or a jewelry store, some sort of purchase of a large ring. All these things are quite possible, so it is very real.
So where is this going today? Well, it's still the basics, right, and in the basics, you must have good data, you must have a place to store your data. This is where I think Oracle can play a role here, but it's not just the data, it's the data and the domain knowledge, and that's the most important thing. You need to know the data, you need to know. It's not just your bonus amount this year, it's your bonus amount this year versus last year compared to your peers. It's the rate of change of the number of opioids that you're taking compared to what you used to be doing. It's all these temporal kinds of data and comparative and derive variables that are very specific. It has nothing to do with machine learning algorithms, but they are the most important thing to get you started.
So, there are, of course, machine learning algorithms, and Oracle has, fortunately, great libraries of all of these. We have about 30 machine learning algorithms that run in the database. We have about 30 machine learning algorithms that run in Spark and Hadoop. You gotta have the data, you gotta have the domain knowledge with the data. Those kind of go together in my mind, the algorithms. And then what does that generate? That generates models, predictions, and insights, and it makes you feel like that, although that's a little bit science fiction-y.
But really, from a more practical point of view, it gives you the ability to hit your customer with the right product at the right time, anticipate things, know what's a healthy outcome, and really have much greater insight into the, I guess, future of your customers.
And so, that's all important, but the most important thing is to operationalize this, because if you don't deploy and operationalize your analytical methodology, you just have a list of customers on a piece of paper. You have an interesting report, you have an interesting pie chart, but you need to deploy that, you need to operationalize that. And if you remember what I said in the beginning about how Oracle brings that algorithms to the data, that changes everything.
I have recorded my talk about how algorithms fuel these changes, real world examples, and a whole lot more. Click on the video below to view it.
If you are interested in how to apply machine learning, algorithms, and Big Data strategies to you own business, visit Oracle Big Data.