The solution of the clustering problem is based on the concept of "similarity" ("proximity") of objects in the feature space. The formalization of this concept is the most consistent with human ideas is the function of competitive similarity (FRiS-function). The effectiveness and applicability of the FRiS function have been proven in solving problems of analyzing data of various types.
Clustering algorithms for space-time events or objects with spatial and temporal characteristics suggest searching for groups of events that occurred “close” from each other in space (in terms of the distance between points, calculated by geographic coordinates) and “close” in time of appearance. To determine the "proximity" of events can be used a priori threshold values. Note that for some events the “natural process” of group formation in time and space can be defined as follows: each next event in time initiates a new group of events or occurs close to an already existing group. At the same time, the spatial "proximity" of events can be considered in a competitive situation, that is, to which of the existing groups is closer a new event. Therefore, for the clustering of space-time events, it seems promising to develop an approach to the taxonomy of data based on the concept of competitive similarity. A modification of the FRiS-Tax algorithm for space-time events is proposed and an example of solving the problem of grouping data on lightning discharges using the proposed version of the algorithm is considered. This clustering method applied for clustering analysis of data on thunderstorm activity using data from the World Wide Lightning Location Network.