ABSTRACT
The areas of social science, environmental monitoring, financial and economic, health and geographic information have generated a large amount of time-varying multivariate data. With a depth analysis of the timevarying multivariate datasets, we find that some of the datasets contain only a numerical dimension (variate) and certain categorical variate, and the numerical variate is the “metric” of the categorical variates, its value indicates the specific quantity of a classification determined by the categorical variates. The numerical variate (attribute) is called “metric properties”, and correspondingly, this type of datasets is called “Time-varying Multivariate Data with Metric properties (TMDM)”. For Example, sales data typically includes the sales of more than one point in time in the economic field. And the data for each point in time, includes multiple variates (type of product, sales locations, sales volume, etc.). Every variate changes over time, and sales volume is the metric property.
