Published on March 18th, 2016 | by Joshua New0
10 Bits: the Data News Hotlist
This week’s list of data news highlights covers March 12-18, 2016 and includes articles about an algorithm that allows self-dirivng cars to nagivgate in bad weather and an analytics platform that can combat illegal fishing activity.
Adobe has launched a tool called the Digital Price Index (DPI) to provide better insight into online price fluctuations using data from billions of transactions conducted with its its online retail software. The U.S. Bureau of Labor Statistics produces the Consumer Price Index to track the price changes of common goods and indicate inflation, however it is not well suited to tracking prices online, which now account for seven percent of U.S. gross domestic product. Adobe’s DPI can provide policymakers and economists with real time data about online pricing to augment official CPI data and provide a greater level of insight into the economy as a whole.
Pierre Duquesnoy, creative director for London-based marketing firm DigitasLBi, and pigeon racing enthusiast Brian Woodhouse have developed a system they call Pigeon Air Patrol that uses air quality sensors carried by pigeons to map levels of pollutants in London. Duquesnoy originally planned to use sensors mounted on drones but chose pigeons instead since it is illegal to fly drones over London. Pigeon Air Patrol, which uses lightweight sensors developed by Paris startup Plume Labs, releases pigeons in different neighborhoods throughout the city and publishes analysis of the data online. London residents can also tweet their location to the Pigeon Air Patrol twitter account to receive live readings of air quality in their area.
A major challenge for self-driving cars is navigating in bad weather, as they rely on LiDAR (a portmanteau of “light” and “radar”) to map the nearby environment using lasers. However, when these lasers hit a raindrop or snowflake, they can divert the lasers and generate confusing “echoes.” To solve this problem, Ford and the University of Michigan have developed an algorithm that analyzes these echoes to determine if the interference was caused by a snowflake and not something else, such as a person standing in the road.
The Taiwan Central Geological Survey has launched a public database of information about areas prone to soil liquefaction, a phenomenon that causes soil to behave like a liquid when shaken, reducing its structural integrity. These areas are particularly at risk for damage during an earthquake, and the Taipei City Government is funding the development of detailed maps of these areas. The database will also provide digital alerts to users during an earthquake. Soil liquefaction is suspected to have been responsible for the collapse of a building and the deaths of 117 people after an earthquake struck Taiwan last month.
Microsoft artificial intelligence researchers have created a software development program called AIX that allows artificial intelligence-controlled characters to explore the open-world video game Minecraft. The goal of AIX is to train artificial intelligence-powered characters to play Minecraft like a human, which entails navigating varied terrain, building structures, and avoiding dangerous elements in the game, by having them learn from their own experiences. Microsoft will make AIX open source this summer to allow anyone to experiment with the platform.
Marine advocacy group Oceana has announced that its system for tracking fishing vessels and identifying illegal fishing activity with satellite data, called Global Fishing Watch, has helped dramatically reduced illegal fishing in the Phoenix Islands Protected Area (PIPA), the largest protected area in the Pacific Ocean. Global Fishing Watch, developed in partnership with Google and environmental mapping nonprofit SkyTruth, monitors location data broadcast by ships to help avoid collisions at sea and automatically flags suspicious behavior that could indicate illicit fishing, such as a ship suddenly turning off its broadcast near protected areas.
Researchers using a supercomputer at the Lawrence Livermore National Laboratory in California have developed a 3D model of all the arteries larger than one millimeter in diameter in the human body that can realistically simulate how blood flows through the cardiovascular system. This model can help researchers study how different treatments for cardiovascular disease, such as a stent or surgery, will impact blood flow. The researchers are working on expanding the simulation to model the other half of the cardiovascular system—veins—and eventually even tracking individual blood cells, which could help researchers simulate how cancer cells or other diseases might spread through the human body.
New methods of data collection, such as a radar tracking system called Statcast and neurological monitoring, are modernizing the baseball analytics system sabermetrics, the analytical technique made famous in the 2011 movie Moneyball. This new data sources can help analysts better model player performance and monitor and reduce injuries. In addition, data about players’ neurological activity could allow analysts to generate new insights into how to improve training programs by better understanding how a player responds to different activities.
The Transportation Department of San Jose, California, is using data analytics and geographic information systems (GIS) to identify sections of roadways responsible for large shares of fatal or serious car accidents and develop strategies to make these roads less dangerous. By identifying factors that contribute to car accidents, the San Jose Police Department and other emergency agencies can prioritize where they station officers and city planners can engineer better infrastructure solutions to reduce the risk of accidents on these roads. The analytics work is part of San Jose’s efforts to implement the Vision Zero initiative, an international effort to eliminate traffic fatalities through education, engineering, and enforcement projects.
Researchers at the University of Rochester have developed a machine learning algorithm that can identify tweets sent by Twitter users while drinking alcohol. The researchers trained their algorithm to identify tweets that contained alcohol-related keywords, such as “drunk,” “beer,” or “party,” and analyzed the tweets’ geotag and timestamp to identify if the user was drinking at home or somewhere else, and at what time. By testing their algorithm on 11,000 geotagged tweets sent in New York City, the researchers were able to make detailed maps of areas with high levels of alcohol-related activity. The researchers plan to expand their analysis to better understand how other factors, such as social setting, age, and ethnicity influence alcohol consumption.
Image: C G P Grey.