Abstract: Competition conditions are increasing rapidly in almost every sector today. Along with the developments in the e-commerce sector, it has been seen that most of the developed countries are integrated and the development of logistics sector increases rapidly. Considering this increase in almost every sector, customer loyalty is of great importance for companies. By taking advantage of the data mining technology and taking into consideration the behavior exhibited by customers, the data obtained can???????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????? be modeled to determine the customers who have a tendency to leave the company. In this study, it was tried to reveal the lost customer behaviors by examining the shipping information of the customers working with a logistics company operating in Turkey. Data about 2.000 customers from the data received from the company were used in our application. Based on the customer shipment information, the input data were created by dividing into 5 classes. In the output data, the acquired and lost customers were taken into consideration. The information obtained by the data mining has been tested on the support vector machine algorithm. The data of these customers pertaining to past two years were divided into 3-month periods. Customer loss analysis was conducted for a total of 8 quarters including 7 sets of input data and 1 set of output data, and it is tried to make loss analysis estimation for the customers who have a tendency to leave the company in the next three months.

Keywords: Churn Analysis, Data Mining, Classification, Support Vector Machine.