Dr. Daniel Nikovski, Group Manager of the Data Analytics Team for the Mitsubishi Electric Research Labs (MERLE) in Cambridge, Massachusetts, explains his work including his collaboration with prominent researchers in Academia on Advanced Primitives on Time Series Analysis providing an explanation of the progression of developing this technology. He also explains the importance, power, and use of this technology.

Video Transcript

Mr. Ted Hill ICONICS President and CEO introduces Dr. Daniel Nikovski Group Manager of the Data Analytics team for the Mitsubishi Electric Research Labs (MERLE). [0:00] 

Dr. Nikovski explains his presentation topic on data analytics technologies and provides an overview of his work and the long and rich history of MERLE. [1:18] 

Dr. Nikovski explains that he works on technologies to make better decisions from data and explains the two main classes of the technologies: predictive modeling and decision and optimization. [3:29] 

Dr. Nikovski begins his presentation on predictive modeling and explains its application. [4:04] 

Dr. Nikovski explains that a good way to compute analytics efficiently on time-series data is to use time-series analysis primitives and provides an explanation. [6:52] 

Dr. Nikovski talks about Shapelets, a type of primitive, and how his work and collaboration Professor Ayman with the University of California at Riverside resulted in finding patterns in the fastest manner possible. [8:06] 

Dr. Nikovski explains the case of prognostics and how to use time-series data to predict failure. [9:37] 

Dr. Nikovski explains other classes of primitives: motifs and discords. [11:11] 

Dr. Nikovski explains a better way to find motifs and discords and explains the matrix profile of time-series data. [12:33] 

Dr. Nikovski explains the significance of the matrix profile and its use in computing motifs and discords. [14:49] 

Dr. Nikovski discusses how the proposal of the matrix profile has enabled the creation of time series chains. He explains what these are and discusses an example. [15:50] 

Dr. Nikovski explains the algorithm used to find time-series chains. [17:51] 

Dr. Nikovski reiterates the importance, power and use of time-series analysis primitives and ends his presentation. [18:57]