Customer Segmentation of the Ports and Maritime Organization using Self Organization Map and K-Means algorithm

Given that organizations' two-way relationships with their customers have changed significantly over the past few years, there is no long-term guarantee of business continuity. Therefore, in order to maintain competitiveness in this uncertain market, organizations should identify their customers well, anticipate their needs and wants, and equip themselves with this information and provide effective marketing strategies to maintain their survival. Given the importance and high share of Iranian port revenues in the domestic economy and the existence of fierce competition between ports in the region, the need to identify key customers and determine their needs and wants for the Ports and Maritime Organization is felt more than ever. On the other hand, data mining, which is the science of data analysis, is introduced as a bridge between parts of data. In this regard, there are tools in data mining such as clustering and classification that create the necessary conditions for the organization to provide the desired service to the customers of the target cluster and to establish a close relationship with them for the organization. Therefore, in this dissertation, RFM analysis is performed on the processed data of 595 customers of the Ports Organization during one year and the clustering process is performed using RFM analysis output and two clustering algorithms, K-means and SOM, in order to determine the optimal number of clusters is used from the silhouette index. The C4.5 tree algorithm is then implemented on the results of the two algorithms K-means and SOM and customer behavioral characteristics are identified. Finally, the quality of the clusters is evaluated using the standard deviation of the data within the clusters and the results obtained from the two methods are compared. Due to the fact that the quality of the clusters obtained from the SOM algorithm is better than k-means, based on the clusters obtained from the SOM algorithm, key and valuable customers are identified. Based on the analysis of the results, it was found that the customers of clusters 9 and 12 resulting from the SOM algorithm with the ↑ M ↑ F ↑ R model have the highest value and loyalty for the ports organization and are the most important customers of the ports organization and the customers of clusters 1 resulting from SOM algorithms with ↓ M ↓ F ↓ R model have the lowest value and loyalty for the port organization. In this way, key and valuable customers are identified.