Clustering of Travel Insurance Cases with K-Modes Algorithm

Abstract

Travel insurance is protection against risks that may occur when a person travels, including tourist trips. Information about the characteristics of tourists helps insurance companies in creating new products. In this study, travelers will be grouped based on the attributes of age category, income category, the number of families category, education level, type of work, history of health, frequency of travel, and frequency of going abroad. Clustering for mixed data nominal and ordinal usually pay less attention to ordinal attribute information. We use the k-modes algorithm with a measure of proximity that positions the existence of the essence of the sequence on the ordinal attribute. We classified 710 tourists who already have travel insurance into two clusters based on this method. At the same time, as many as 1277 travelers who do not have travel insurance were into four groups. Based on the profiles of each group, we conclude that there are similarities in the characteristics of 290 travelers who do not have travel insurance with 332 travelers who already have insurance. This group is private workers who graduated from college, are 30 years old, have no history of chronic disease, have a family of four, and are upper-middle-income. This group also rarely travels and never abroad. Insurance companies can target prospective tourists with these characteristics in offering their products.