mercredi 24 octobre 2012

Nowcasting: Big data predicts the present

A lire sur:

By Irfan Khan,

Source: Cea./Flickr
Big data is the key ingredient for predicting the present, or nowcasting, as it's called. Conceptually, it seems silly to use resources to predict the present. After all, we all live in the present and can see what's happening around us. But it's not silly at all. Nowcasting can save people's lives and make companies money.
Nowcasting is a term used among meteorologists to forecast near-term weather conditions, particularly violent weather approaching a particular region, based on vast amounts of incoming data. Knowing that a tornado is minutes away from your neighborhood can be the difference in whether your family survives or not.
Thanks to the pervasiveness of big data, nowcasting is useful beyond weather. For example, the European Central Bank has published a paper on how it uses nowcasting of huge volumes of economic data to get an early jump on GDP estimates, which are critical to establishing short-term and long-term fiscal policy.
Hal Varian, Google's chief economist, suggests that nowcasting can be used to predict consumer behavior in specific regions. Varian and Google economist Hyunyoung Choi wrote in a paper that by using Google Trends data for a given geography, highly accurate predictions can be made for such major economic activities as car and home purchases. By modeling search data for, say, "real estate agents" in Texas, they are able to more accurately estimate home sale activity in the region.
Their research has been validated elsewhere. For example, economists in Hungary have followed Varian and Choi's nowcasting methodology, and created extremely refined estimates of consumer activity in that country.
Nowcasting is becoming a valuable tool for anyone interested in economic and business conditions in the present. And knowing with better precision what's happening now will give us a better understanding of what might be in store for the future.

Aucun commentaire:

Publier un commentaire