Использование метода главных компонент для анализа надежности цепей поставок

Логистика и управление цепями поставок. 2018. № 4 (87). С. 27–33.

Kuznetsov V. O.

One of the options for a more flexible approach to analyzing the reliability of supply chains is the principal component analysis (PCA). With a large number of variables describing supply chain, it is a difficult task to analyze the structure of variables in two-dimensional space. Within the analysis of the variables dependencies PCA allows to go from multidimensional space to low-dimensional space, leaving the most informative data in the array for analysis. Based on the generated data set, this paper demonstrates a possibility of applying PCA to supply chain reliability analysis. The generated data set includes observations of 50 supply chains described by five variables. Based on the array, maximizing the linear combination of parameters for each observation, we determined load coefficients and estimates of each of the main components. The calculation of these coefficients made it possible to move from multidimensional space to a two-dimensional one. The two-dimensional representation of all the data whose axes are the first two main components, explaining 84% of the variance, allows to see the structure of all supply chains, to identify outsiders and leaders in this set.

Keywords: машинное обучение, надежность, цепь поставок, метод главных компонент, supply chain, «обучение без учителя», machine learning, Unsupervised learning, principal component analysis, Reliability