Preferred Infrastructure, Inc.

Concept

Jubatus is a “large-scale distributed real-time machine learning platform” that analyzes big data in real-time.

Computer data used by mankind is increasing continuously at an exponential rate. Today, this kind of large-scale data known commonly as “big data” is growing in various fields. Moreover, this does not indicate that the amount of data is simply just increasing. Big data usually consists mostly of unstructured data, log files, sensor data and other raw data that did not need to be processed previously. Recently, it has become possible to find useful information from large-scale data from the perspective of computational power and data analysis technology.

In the current big data analysis era, it is important to be able to analyze data efficiently and swiftly lead to further processing. Jubatus makes this possible as an analysis processing platform.

Jubatus has 3 main concepts. Real-time functionality, large-scale processing, and deep analysis – these 3 key points are emphasized to work together. In previous data analyzing technologies, only 2 of the mentioned points could be achieved at the same time. Large-scale distributed batch processing platforms has made large-scale processing and deep analysis possible. Similarly, online machine learning methods have made real-time functionality and deep analysis possible. Finally, streaming technology has made real-time functionality and large-scale processing possible. These 3 key points make up Jubatus and are what Jubatus emphasizes more than anything.