System analysis: the kemeny’s median as an optimization model of the preferences group system of air traffic controllers of the danger of the characteristic errors

Abstract

The systems of advantages of aviation operators of the “front line” on the indicators and characteristics of their professional activities is one of the indicators demonstrating the influence of the human factor on decision-making, and, consequently, on the “attitude towards dangerous actions or conditions”, which, in its turn, is one of the components of the current ICAO safety paradigm. The preference system is understood as an ordered series of the specified indicators and characteristics from the most dangerous to the least dangerous, including errors that can be made by air traffic controllers. Group systems of advantages have a number of properties (peculiarities of the prevailing in a particular society — control shift — opinions on the perception of threats and hazards, the influence of the attitude of instructor personnel to threats and dangers and the technology to overcome them, statistics of aviation accidents and serious incidents in the area of responsibility, etc.) that are desirable to take into account in the safety management process and that are found by aggregating individual systems of benefits. This aggregation occurs using strategies for making group decisions, from which one should point to the strategy of summing and averaging ranks, which is riskier, but allows establishing the degree of consistency of opinions using the Kendall concordance coefficient. An important strategy is based on the classical Savage decision-making criteria, which has an optimization content and allows minimizing deviations in opinions regarding the dangers of mistakes of both the majority and the minority of the group members. The Kemeny’s median has a pronounced nonparametric optimization content, but it is almost never used in studies of the influence of the human factor on decision making in aviation systems. Individual systems of preferences of m=37 air traffic controllers on the spectrum of n=21 characteristic errors were constructed by them using the usual method of pairwise comparisons and normative establishment of a part of the total hazard. The use of the technology for detecting and filtering out marginal thoughts — individual systems of advantages, which significantly differ from the general group, made it possible to distinguish a subgroup mA=26 with a high level of intragroup consistency of opinions: the coefficient of concordance is W=0.7144 and is statistically significant at a high level of significance a=1%. Individual preference systems of members of the mА subgroup were used to implement the heuristic algorithm and construct the desired Kemeny’s median, which improves the consistent preference system and has an unusually high coincidence with the group advantage systems obtained using other group decision strategies: the average value of Spearman’s rank correlation coefficient in 7 times increased its minimum acceptable value.