An approach to automating the daily performance indicators monitoring of a transport company

Abstract

The article compares the results of applying some statistical criteria to detect suspicious values (gross outliers) in the daily performance indicators of one of the transport companies in the Russian Federation. The criteria applicable to data samples with a normal distribution law are considered. In this connection, a preliminary test of the hypothesis about the correspondence of the distribution of controlled quantities to this law was carried out. It is shown that identifying gross outliers in the values of the indicators under consideration requires the simultaneous application of several statistical criteria.