การพัฒนาคลังข้อมูลลูกค้าและใช้เทคนิคการทําเหมืองข้อมูลเพื่อการบริหารลูกค้าสัมพันธ์ : กรณีศึกษาโรงพยาบาลแมคคอรมิคจังหวัดเชียงใหม่

Abstract

This study involves research and development. The purpose of the research was to develop a customer data warehouse using data mining techniques for customer relationship management. A data warehouse was developed and analyzed from the outpatient data of all departments from January 2011 to December 2013. The research process consisted of two parts : 1) development of the data warehouse 2) mining the data warehouse.

The development of the customer data warehouse used Microsoft SQL Server 2000. The data warehouse model used a star schema. For loading data into the data warehouse, we used Microsoft Visual Studio 2005 in order to create an ETL process for data cleansing, and formatting, and for the importing of 696,457 records. The data mining used clustering techniques to group customer and time series techniques to predict numbers of customers.

The clustering technique, which grouped customers into 9 clusters according to their characteristics, and a group of customers who used the service are the cluster 7, who use the service vaccine department. A group of customers has 109,111 records and accounted for 15.7 percent of all data. Resulted in new knowledge about customers. Executives of McCormick Hospital could use the results for supporting decision making, marketing promotion and activity promotion for each customer group to manage customer relationships. Evaluation of the results is at 0.998, which means that the measurement tools were reliable. The time series forecasting the number of customers who used our services. Most customers used the service medicine department. In January 2557 a number of customers was increasing. On Thursday had many customer used the service at 10.00 am and 07.00 pm. As for the time series technique which predicted the number of customers by comparison with the actual data, they were nearly the same number and the trend was in a similar direction.