This week’s list of data news highlights covers September 26 – October 2, 2015 and includes articles about the White House’s efforts to promote citizen science and an algorithm that can help in the fight against ISIS.
The U.S. Department of Transportation is building the first publicly accessible database of the geographic location of every address in the United States. Though other agencies such as the Postal Service and Census Bureau and state and local governments maintain address databases, rules restricting public access and the patchwork nature of these systems have historically limited the value of this data. The project will evaluate data the federal government already has and work with state and local governments to collect missing data to develop a publicly accessible database that could help the government reduce wasteful or redundant efforts to collect and manage address data.
The National Aeronautics and Space Administration is working on a partnership with the U.S. Agency for International Development (USAID) to use satellite imagery and data analysis to help developing countries prepare for natural disasters and respond to the effects of climate change. The partnership, called Servir, allows USAID to manage resources more effectively, monitor crises on a more granular level, and carry out live-saving crisis preparation activities. For example, USAID used Servir tools to predict flooding patterns eight days ahead of dangerous flooding in Bangladesh, reducing total flood casualties in the country during last monsoon season from the expected several thousand to just seventeen.
Federal Bureau of Investigation (FBI) director James Comey has announced that the FBI will collect more data about law enforcement shootings and will publicly report on the police’s use of deadly force. The FBI’s report will detail shooting incidents, the parties involved, the nature of resulting injuries or deaths, and the relevant circumstances of each incident. Comey described the current lack of comprehensive data on police shootings as a hindrance to meaningful discussion of the issue.
The Associated Press, an international nonprofit news agency, has received a grant from the Knight Foundation, a nonprofit that fosters innovation in journalism, to expand its data-driven journalism efforts. With the grant funding, the Associated Press will establish best practices for data journalism in its AP Stylebook, a widely-used guide for journalism standards.
The White House Office of Science and Technology Policy (OSTP) has taken steps to promote the use of citizen science (when volunteers contribute scientific data, analysis, and technology for public benefit) and crowdsourcing by government agencies. OSTP released a memorandum directing agencies to leverage these activities to augment government capacity, as well as make it easier for the public to participate in these efforts. OSTP also released the Federal Crowdsourcing and Citizen Science Toolkit to provide agencies with best practices for designing and implementing citizen science and crowdsourcing projects.
Hitachi has developed an analytics system called Hitachi Visualization Predictive Crime Analytics (PCA), which is capable of predicting where and when crime will likely occur based on a wide variety of data, including connected sensors, social media, crime statistics, and weather reports. PCA maps cities with color-coded indicators of crimes and can assign threat levels to specific locations, which Hitachi says can guide law-enforcement decision making. Several cities, yet to be announced, will pilot PCA in October.
Researchers from Arizona State University have developed an algorithm that can model the behavior of the extremist group, ISIS. The researchers developed the algorithm, which they describe as a proof of concept, by applying machine learning techniques to 2014 data about the behavior of ISIS. They were able to identify insights such as how a spike in car bombs in Baghdad typically occurred before ISIS attacks in northern Iraqi cities, suggesting that ISIS planned the car bombs to divert Iraqi security forces. The researchers hope that with additional and real-time data from the Department of Defense, their algorithm could develop more valuable insights.
Researchers at Cornell University and Stanford University working on a project called Brain4Cars have developed software that can predict when a driver will change lanes with 90 percent accuracy by analyzing how drivers move their heads while they drive. The researchers trained deep-learning algorithms with data collected on 10 drivers driving a total of 1,180 miles, focusing on telltale head movements preceding a lane change, such as a glance over the shoulder, as well as changes to steering, braking, and acceleration. Such behavior prediction algorithms could make automatic braking and steering systems, common in luxury cars, more intelligent and safer.
The 1000 Genomes Project Consortium, an international public-private scientific research consortium, has completed its goal of sequencing the genetic variants of a massive amount of people to develop a comprehensive resource on human genetic variation. The largest-ever database of its kind contains analysis of 88 million variable sites on the human genome, taken from samples of 26 representative populations from Africa, the Americas, Europe, South Asia, and East Asia. The database will serve as a resource for scientists evaluating the genomic variants associated with various diseases, removing the need to sequence entire genomes themselves, which can be costly.
Airplane manufacturing company Boeing has announced a new partnership with Carnegie Mellon University to establish the Aerospace Data Analytics Lab to research how artificial intelligence and machine learning can improve aircraft manufacturing and plane safety. The lab will analyze data on airplane design, construction, and operation to develop new insights into processes that can affect safety, such as maintenance schedules. Boeing’s initial investment will fund the lab for three years.
Image: Staff Sergeant Val Gempis.