Massachusetts Institute of Technology (MIT) associate professor Devavrat Shah has announced the creation of a new algorithm that can predict Twitter trends hours in advance.
The new algorithm was created by Professor Shah and MIT graduate student Stanislav Nikolov. The MIT alums say the new algorithm can predict trends with 95 per cent accuracy. According to Shah, the algorithm can even predict trends up to five hours before Twitter announces them.
The algorithm works by comparing past data to real-time information in a new way. Traditionally, algorithms compare similar old and new data equally.
However, Shah and Nikolov's new algorithm compares older data on a curve. The algorithm assigns a higher value to previous data that follows a similar trajectory to the present data being analyzed.
In essence, the algorithm compares old and new data in real time. So the algorithm can give a probable likelihood that a subject will trend based on old data that has been weighted in comparison to the subject.
Shah and Nikolov used a sampling of 200 subjects that did trend and 200 subjects that didn't trend to test their algorithm. The team found that the algorithm offered 95 per cent accuracy with a four per cent false-positive rate.
In a press release on the discovery Shah says the new algorithm could be used by Twitter to charge marketers for ads linked with impending trends. Which means someday you could see Twitter ads for Ford cars before you even thought you wanted to tweet about autonomous driving vehicles.
Indonesian overclocker Ivan Cupa (with the aid of a lot of liquid nitrogen) achieves record overclock on AMD's latest Threadripper
Ssupermassive black hole is so big it corresponds to four per cent of the galaxy's total mass
Imminent attack will target a single bank with cloned cards used to fraudulently withdraw millions over one weekend
Using photocatalysts to convert carbon dioxide into usable energy such as methane or ethane