Predicting the Future

In This Article:

- How data analysis helps creating a better and smarter planet. 

  

On some fantasy worlds  some magicians can predict the future. In "Lord of the Rings" for instance some elvish elders posses this capability. Interestingly, in a sense, predicting the future is what physicists dream of; finding the world formula! 

A world formula is mathematical description of all the physical forces in the universe combined in one equation. Assuming we had such a tool, we still needed to know the so called "boundary conditions" of every particle, from Galaxies and planets down to molecules and individual atoms, meaning their position, size, velocity, energy etc. In other words we needed to have a lot of data about everything. An this is the catch phrase for this blog - data.

Data availability - the new commodity

    

Now, let's take look what what the data-situation nowadays is like. So, the internet traffic is growing with about 50% every year (CAGR), that is significantly faster than the all-so-fast-growing-doubling-of-transistors-on-a-Intel-chip aka. "Moore's Law". Now, the internet growth is certainly not the only data-growth engine, but with the trend of cloud-computing, all data is virtually going over the net and not just sitting in local data-servers. Or try this - as of 2011 there are 2 Billion people on the internet, 4 Billion mobile subscribers and more than 1 Trillion that is 1,000,000,000,000 electronic devices like the smart phones connected to the world wide web creating and sharing massive amounts of data. That with so much data, there could be an energy and power issue related to such data traffic is foreseeable, but let's leave that for another post. 

Now, why does the data rate grow so fast? Well, there are many reasons - try social networking, smart phones, sensors for environmental and social surveillance etc. With so much data at hand we can think about using it for a smarter and better planet. How? Let me give you an example from a talk I recently heard given by Bernie Meyerson the IBM VP for Innovation (you can watch the talk here for free). He starts off with the the key word transportation WRT urbanization. Since of 2010 more humans lived in cities than in rural areas - trend fast growing. Thus, challenges like food, water, health-care and transportation for these becoming mega-cities are growing. Per exemplum, let's take a look into car traffic and the resulting traffic jams, and what smart data mining can do to solve it. 

Analytics to save the day

Singapore is a suitable example for such traffic issues. However Singapore was lucky enough to have an  traffic monitoring system installed, where the car density can be tracked in real-time. Knowing the traffic in real-time is not good enough Meyerson says, since what is it worth to know that there is a traffic jam now (my real time GPS tells me that) if I'm stuck in it - well guess what, I can see the car in front of my bumper very well. So, IBM goes a step further and shows how analyzing such traffic data can be used to predict the traffic density and location of traffic 'hot-spots' for the next 20 or 30 minutes.

But this is not enough: Now, they deploy an mathematical algorithm to the taken data towards extracting valuable information. Thus, not only can we predict the future now, we can utilize these results to change the future of a system (e.g. car traffic) in such a positive way, such that the traffic jam does not even occur! How? For instance, by installing a traffic lighting system, that slows cars at specific locations down such that the critical car-density creating a jam does not occur. In more general words: We are foreseeing and changing the future! ...well at least for 30 minutes and for cars. And it can be easily envisioned how such proper data analysis can be deployed in other topics of global significance such as Health Care, Transportation, Energy, Education, Security etc. 

Want an example for security? Imagine we mark burglary crimes in a city on that cities map. What evolves is a, more or less, complex pattern of such crime acts. Analyzing these occurrences with an smart algorithm allows to predicted  the next crimes place, and potentially time such that preventative measures can be taken. A successful demonstration of this novel tool was implemented in a city scoring within the top 10 crime burdened cities in the US, which dropped to place 200 and recently off that list after running the pilot project - what a better world it must be for those city's citizens.