Real-time Analytics: It’s Always Decision Time!
Greetings all! I just came across a great video from eWEEK talking about the growing need for real-time (aka streaming) analytics:
“For years, business intelligence has provided valuable information to help executives and managers make decisions to increase sales, improve operations, and seize new business opportunities. With the quickening pace of business today and the need to make faster decisions based on more timely data, companies are complementing this data using information mined from social networks, mobile sensors, and even location-based information from smartphones. To get the best value from this wealth of new data sources, the data analysis must be done in real time. This allows decisions to be made based on the true conditions at that particular time.”
PatternBuilders Analytics Framework is designed to handle big data and real-time analytics (see Terence’s post about our real-time/streaming architecture) and while we’ve posted about the benefits of real-time before, Ashley Daley’s video says it all in less than four minutes! In particular, Daley focuses on data velocity:
- How fast does data pour in?
- How fast does its value diminish?
Both questions are important because you need to ensure that your analytics solution can handle the data streaming in as well as provide the appropriate time windows for data capture and analysis before the analysis is rendered meaningless. Case in point: credit card fraud detection (my favorite example). Regular readers of our blog may remember my post on Capital One’s credit card detection capabilities and how fast they were able to detect a fraudulent charge on my credit card. Well, Capital One strikes again and is an excellent example of data velocity.
But before I begin, a little housekeeping: First, I have no relationship with Capital One other than the fact that I have a credit card with them. Second, although I have no business relationship with them, I am a great admirer (first hand) of their fraud detection capabilities and would love to learn more (whenever I have a fraud “incident” I probably stay on the line with them far longer than any of their other customers!).
So, I am at my regular gas station, about to pump my gas, when I swipe my credit card and mistakenly enter the wrong zip code (strike 1). My card is denied with one of those little kiosk messages to “please see the gas station attendant.” I go in to see the attendant, he then swipes my card and asks the amount to put on it and I reply that I want to fill my tank so he puts $75 on the card (strikes 2 and 3). My card is then denied and this time the attendant informs me that he cannot put through the card. I sigh, give him some cash, fill my tank, and leave the gas station at 11:18 (I have a digital clock on my dashboard and the time is very important). I get home, download my email, and see an alert from Capital One fraud detection asking me to contact them immediately. The timestamp is 11:17.
Of course, I call Capital One and this is what I discover:
- In my local area it appears that credit cards have been used to fill gas tanks at multiple stations up to $500 and within hours of each other (they believe that card numbers have been stolen and put in use by multiple people).
- When I entered the wrong zip code, I triggered a first level flag. When I went to see the attendant, one more flag was entered because whenever I’ve pumped gas at this particular station it was always done from the kiosk and then when the attendant opened a charge with $75, my card was put on hold until I contacted them.
- Since my behavior was atypical (hey, I am a creature of habit) and they were seeing trending gas station fraudulent activity, they suspended my card.
How is this an example of data velocity? (I am so glad you asked.) Let’s consider the two questions posed at the beginning of this post:
- How fast does the data pour in? Well, there are $2.5 trillion in transactions at more than 24 million locations in 200 countries and territories and it’s estimated that 10,000 payment transactions occur every second around the world.
- How fast does its value diminish? It would be safe to say in the case of credit card fraud, that every minute counts as demonstrated by Capital One’s hold on my card that occurred less than 60 seconds after my kiosk denial.
Whether you call it real-time or streaming analytics, I think that you get my point. If you are able to analyze the data as it comes in and react to it, it follows that you can change the outcome. In the case of my Capital One card denial, I behaved atypically which was then coupled with fraudulent activity at my local area gas stations. Capital One suspended my card, alerted me immediately, and by 11:23 (digital clock on my laptop) my card was reactivated. How’s that for real-time big data analytics in action? Pretty cool in my opinion!
So if you have a chance, take a look at the eWEEK video (it’s short and sweet), consider how real-time analytics could positively impact your business, and let us know how we can help. And if you want to dive a little deeper, here are some of our others posts on the subject:
- How “Real” is Real-Time and What the Heck is Streaming Analytics?
- It’s About Time: Series Data, Streaming, & Architecture
- MongoSF: Our Streaming Analytics Video is Now Available
- Fraud Detection By The Numbers
Until next time folks! And look for some more posts from us on real-time analytics and why everyone (from the executive suite to the feet on the street) should be able to run them.