The New York Fire Department will start a "data mining plan" to predict a fire

ByMarines

New York City, the largest city in the United States, has more than 1 million buildings, of which approximately 3,000 households have experienced major fire every year. As a countermeasure to that, the New York City Fire DepartmentData MiningWe are planning to exploit the high firing rate building by utilizing the method, and to start an attempt to prevent fire.

How New York's Fire Department Uses Data Mining - Digits - WSJ
http://blogs.wsj.com/digits/2014/01/24/how-new-yorks-fire-department-uses-data-mining/


The risk of fire occurrence varies depending on the characteristic factors of the building. Jeff Chen, an analyst at the New York Fire Department, took place in Las VegasConference on Big Data Utilization conferences"We are correlating with the" low-income neighbor "and the occurrence of the fire" announced the analysis result.

These factors are determined by data mining from huge data possessed by New York City and Fire Department, and other factors include "Number of years of building", "Number of installed sprinklers", "presence of elevator", and so on. "For the vacant houses and unmanned buildings the fire rate doubles," Chen said.

ByEmpty Quarter

So New York City built an algorithm to determine the "risk score" that can be used for the examination of approximately 330,000 buildings excluding single house and two-family house in New York from about 60 different factors. We have created a priority inspection list that collects buildings with high fire occurrence rates and decided to incorporate them into building inspections that fire department officials conduct weekly. Compared with the conventional choice of relatively random buildings, more effective fire checking effect is expected.

ByU.S. Army Garrison Yongsan

This data mining plan was implemented from July 2014, and after that it is planned to extend the factor of building algorithms to 2,400 over several months. However, as the plan succeeds and the "fire that was supposed to have occurred" does not occur, there is a possibility that the algorithm change process becomes difficult.

Jeff Ross, deputy general manager of the New York Fire Department said, "The threat of fire should be prevented beforehand, we will watch the fire rate fall first."