The methodological approaches to carry out expenditure-based distribution centers in logistics service And e-commerce

The methodological approaches to carry out expenditure-based segmentation include two broad categories: (i) segmentation based on threshold values; (ii) segmentation based on clustering analysis. The first approach uses thresholds such as quantiles (e.g., quartile) of the distribution of expenditure variable as the segmentation method. The second approach uses clustering analysis (typically, K-means) to segment the objects based on their volume of expenditures. While the threshold-based approach is somewhat arbitrary, the clustering approach does not provide necessary statistics for inference and fails to account for unobserved heterogeneity. The applications of finite mixture models (FMM) to model the probability of belonging to each unobserved group (i.e., an expenditure segment) and to estimate distinct parameters of regression models are, however, scarce in the literature on market segmentation (Mortazavi & Lundberg, 2019). FMMs can assign the probability of belonging to each expenditure segments based on the level of spending and allows statistical inferences to be made regarding segmentation effectiveness. This paper uses FMM approach and identifies unobserved groups of freight transport expenditure patterns among establishments through a probabilistic framework