Drawing and interpreting the decision tree of the border market Inche Barun customers using a hybrid approach

Aim and introduction: One of the main players of border bazaars whose existence and prosperity is tied to their presence are the customers whose buying behavior has been less noticed in the shadow of other important aspects of border bazaars. In this research, the customer journey map theory was used in order to identify the indicators affecting the customer's shopping journey. Customer journey maps include online and offline interactions and pre- and post-purchase encounters and are useful for gathering insight into how a set of experiences unfolds over time through sequential, service-based processes. The increasing attention to customer journeys shows management interest in increasing value through improving customer experiences.
Methodology: The purpose of the current research is to design a decision-making model for the customers of Inche Barun border market using a hybrid approach, so this research is applied, in terms of descriptive-survey purpose and in the form of cross-sectional studies, which was carried out in two related steps. The statistical population of the research in the qualitative part included the visitors and customers of border market, which was conducted using the theoretical sampling method until theoretical saturation was reached in a total of 21 interviews. The statistical population of the research was in the quantitative part of the visitors of border market, and 389 questionnaires were found to be usable after cleaning and preliminary screening, and the analysis was done on them. In the first step, using the content analysis method, 34 indicators affecting the purchasing journey of customers were extracted and based on that, a research questionnaire was designed, and its reliability and validity were checked and confirmed using Cronbach's alpha and confirmatory factor analysis, respectively. In the second step, quantitative approaches (clustering and decision tree) were used to investigate the statistical population. In order to describe the research population, 389 visitors of Inche Barun market were grouped and analyzed in three separate clusters using Davis-Bouldin index. A decision tree was used in order to draw the pattern of customers' decision-making during the shopping trip to the bazaar.
Finding: The results of the research in the qualitative stage led to the identification of 34 concepts, and the classification of these concepts according to their semantic and conceptual affinity led to the identification of 6 main categories, which, considering that these concepts and categories represent the wishes and expectations of customers in their language, It will be very helpful for officials and decision-makers in this area. The application of decision tree showed that by using the indicators affecting the buying behavior of customers, it is possible to predict their return rate in the first, second and third clusters with the accuracy of 0.851, 0.826 and 0.852 respectively. This shows that the decision-making models among the three clusters do not differ much in terms of accuracy, but the analysis of the customers' decision-making patterns shows that the customers' criteria for repeating the shopping trip are very different, so that in the first, second and third clusters, the index that The zero nodes are, respectively, the index of sanitary ware, trash can, drinking water, product prices and discounts, and road lighting facilities.
Discussion and coclusion: The results of the research in cluster analysis showed that customers touch different contact points during their shopping trip to border market and based on these points, different clusters can be extracted, which shows that the statistical population under investigation is heterogeneous, as well as using the decision tree model to draw the decision pattern It showed that the buying behavior of customers is complex, hierarchical and multi-criteria. The clustering of border market customers shows that although customers touch different and diverse contact points in their shopping journey, the importance of these points is not at the same level and extent for all travelers, and many points are not at the center of people's decision-making, which separates the sample. Statistics in three clusters confirm this issue. The analysis of the statistical sample of the current research based on 34 indicators that influence the customer's purchase journey shows that the target population of the cross-border market is not homogenous and the same, and presenting a single and identical program will not be a solution for all of them and will not cover all tastes and interests. In other words, emphasizing a single version for the entire market may work for estimating the demands of one cluster, but the opposite result may be obtained in another cluster. The results of drawing and interpreting the optimal decision tree of customers by cluster analysis showed that in the first cluster, the index of sanitary napkins, trash cans, and drinking water, in the second cluster, the index of product prices and discounts, and in the third cluster, the index of road lighting facilities played a significant role in repeat customer purchases and the zero node of the decision tree of each cluster is assigned to them.