Japan uses AI to estimate the cherry blossom period


In addition to being useful in commercial markets, big data and artificial intelligence can also help accurately predict the flowering period of cherry blossoms. A Japanese company made use of big data and AI artificial wisdom to estimate the six stages of cherry blossom flowering status, and the number of flowering prediction sites increased from 137 in 2017 to 1 thousand.

Shimadzu Commercial Systems Co., Ltd. of Kyoto City, Japan, announced on March 16th that the number of cherry blossom flowering sites provided by the weather information website “Weather ☆ JAPAN” (お天お☆JAPAN) increased from 137 in 2017 to 1,000.

The company fed big data to the artificial intelligence system to construct a flowering estimation module for 1,000 flowering sites. In the past, the system only provided two stages of information such as “flowering” and “full opening,” and this year it will be divided into 6 The stage accurately stated that there were “flowering,” “flowering 50%,” “flowering 70%,” “full open,” “beginning to fall,” and “end.” These data will be updated 4 times a day, 4 times more than the entire season in 2017, and the frequency of updates will increase significantly. This will make it easier for people viewing Sakura to have accurate information before they leave home.

The company spokesperson further stated that the average error of the computer system in 2017 over the past two days is 50%, and the new system can increase the accuracy of the estimated cherry blossoms to 75%.

In terms of actual data, they predict that Japanese cherry blossoms will bloom earlier than in previous years in 2018. Tokyo blossomed on March 21 as early as the 5th of last year, Osaka blossomed on March 23 as early as the 5th of last year, and Aomori blossomed at the latest in April. Flowering on the 23rd (above the forecast on March 15th statistics, the statistical region does not include Okinawa and Hokkaido).

source:  www.itmedia.co.jp

logo of yuyi global technology

YUYI Global Technology