We are getting used to the idea of big data now that Waze knows to send us a coffee coupon as we drive near a Dunkin' Donuts. Analytics is becoming a fact of life. We know pop-up ads will match our interests.
But what happens when we want to analyze data that we haven't collected yet? This week, we hear from a Vencore computer scientist who tackled just that problem. He wanted to solve a problem with analytics, but there was no data to help him…
Tara Grabowsky MD
Chief Medical Officer
For more information, visit HVHprecision.com
I took my first dive into the world of big data during my time at the University of Pennsylvania. Charged with using our experience in engineering to solve a real world problem, my senior design team and I formulated our base question: is it possible to generate a map of any building we enter without prior knowledge of that building? The immediate answer was "no." You can't draw a map without any information about the area of interest. How could North America have been mapped without sailing the coastline? We decided to ignore the nagging naysayers who claimed "you can't build something from nothing!"
We wanted to create a Google Maps of sorts, offering newcomers the ability to find their way around the inside of the building. We needed a system that could be applied across a multitude of interior spaces with fast results. We needed to gather information about a building and then compile it into a spatially accurate map. We needed to know the difference between halls, doors, walls and rooms. We needed to know distances and dimensions. We needed data.
So we turned to the greatest source of loose data that exists: people. Whether aware of it or not, people are data goldmines, acting as constant streams of information output. And with today’s prevalence of mobile devices, people have become constant streams of readable information output.
With this in mind, we began to develop an ad hoc network of mobile devices constantly locating themselves within a building and sharing the data with a home node. Our system worked like this: A user enters a building and begins automatically locating wireless routers via a WiFi-enabled mobile device. The user is then pinpointed to a location based on the triangulation of WiFi signal strengths measured between the mobile device and three different routers. As the user moves throughout the building, the mobile device constantly ‘locates’ itself in relation to the routers present in the building. This individual user location information is then sent to a home node, which can compile the information of many users and overlay them into a heat map of movement. Enough users over a period of time allows the home node to infer the location of halls, walls, doors, and rooms and create a 2-dimensional mapping of the building’s interior.
It was by no means an easy task, but by the end of the year we had developed our mapping system. It was a gratifying accomplishment, but the truly lasting aspect of the project was the versatility we witnessed in the use of data to accomplish a goal. We were able to create a means of data collection, collect and study the necessary data, and apply it to a larger system to produce something creative and useful. We had, in fact, built something from nothing. We had taken a problem for which we had no data and developed a fully-functional method of mapping unknown areas, all thanks to data analytics.
Vencore Data Scientist