Capacity Method of Rare Events Analysis in the Area of Services

Imagine that you are owner of some service. You need to determine for a certain future period the work plan of your craftsmen, the number of consumables needed. To do this, you need to make a forecast of future services number. Classical mathematical methods of working with time series are not suitable for this task. Aggregation of data on services by months and the compilation of a time series can only confuse. Forecasting services should be performed using methods designed to work with rare events. Rare events are devoted to relatively few works. Methods for the study of rare events are significantly less than methods for analyzing frequent events (time series). The most popular method of studying rare events at the moment is the use of the theory of random processes, which uses a stream of Poisson or Erlang events. However, using random streams, one cannot predict the very moment of the occurrence of an event. The paper describes an approach to the rare events analysis, which is based on: dividing events by identifiers of the sources in which they are formed; regression process parameters occurring within the sources, resulting in these events formation; search by any known method of parameters change patterns; the process start itself to obtain a forecast of the following events time occurrence. For the consumption processes and the disturbances growth process, which are the most common processes of the events formation in the economy, a method is proposed for restoring the consumption or accumulating disturbances rate from the rare events history. Services as can be modeled as the process of accumulating disturbances to a certain level. The article is devoted to the application of the capacity method of rare events analysis on real data in the service sector (haircut in a hairdresser, a manicure in a beauty salon, cellular communication services). The task is to restore the function that leads to the acquisition of services, and then predict the following events.