Read the following article and write a summary, agreements, and disagreements with justification. Do not assume that everything in the article is correct. Try to read from a different viewpoint from the author.

Discovering customer value for marketing systems: an empirical

Data mining technologies have been employed in a variety of business managements for discovering useful commercial knowledge or marketing model for many years. Hence, the major marketing issue for airlines is to identify and analyse valuable air travellers recently, so that airlines can attract them for enhancing the profits and growth rates. However, growth rates are always an important issue for airline industries. An empirical case of air travellers markets in Taiwan is implemented in this research. This research proposes a model (FSLC model, RFM model based) via the data mining technologies to discover valuable travellers for airlines. This study partitions the market of air travellers in Taiwan, and the paper generates useful association rules to find an optimised target market for dynamic marketing or CRM systems. Nevertheless, the results of this research can be applied on marketing or CRM systems of the airline industry for identifying valuable travellers. Finally, the purpose of this research is to find high-value markets for marketing or CRM systems of airlines in Taiwan, and the framework can be applied to other industries as well. Keywords: data mining; FSLC model; RFM model; association rules; marketing systems; CRM systems 1. Introduction Data mining technologies have been widely used in science and management fields for many years. Hence, businesses use the data mining technologies to analyse shopping behaviours for the purpose of determining customer values and enhancing customers loyalties (Berry and Linoff 2004; Cios et al. 2007; Chiang 2014). Enterprises can develop specific models of customer value in accordance with their operations (Chiang 2012). Therefore, this research proposes a model to find true customer values of airlines. The method of this study consists of a proposed model, clustering method and supervised association rule algorithm. Data mining usually consists of classification, prediction, estimation, association, clustering and visualisation. Tan, Steinbach, and Kumar (2006) defined the data mining as Data mining is a process to discover useful information in large data repositories automatically; Linoff and Berry (2002) defined the data mining as Using data analysis and machine learning methods to process data and to create useful models. However, the technologies of the data mining are widely used in marketing or CRM systems to find customers knowledge and it can enhance their contributions and loyalties as well. Businesses implement a variety of marketing plans for each market. The plans are to enhance shopping frequencies, to promote high profit/expense products and also to create long-term customers. Customer values are usually estimated by RFM (Recency, Frequency and Monetary) model as well as for increasing profits of business. Typically, the RFM variables are profit variables for businesses (Drucker 2003; Berry and Linoff 2004; Chiang 2014). Thus, this research replaces the RFM variables with the other profit variables for features of the air travel market in Taiwan. In the tourism industry, international airfare is always a large ratio of transportation cost for international traveller. Therefore, Low-Cost Carrier (LCC) attracts many international travellers with low price. In Taiwan, two new LCC airlines were established in 2013: Tigerair Taiwan and V Air. In 2014, the two new LCC airlines started to operate. That is, the short and medium airline routes of international airline markets in Taiwan are highly competitive between regular and LCC airlines. Hence, marketing or CRM systems are important marketing tools for increasing growth rate of airlines. However, how to estimate customer value efficiently is the main step for discovering valuable and profitable customers in marketing or CRM systems (Wilson, Daniel, and McDonald 2002). The purpose of this study is to employ data mining to analyse customer values of international travellers for enhancing growth rates of regular airlines. Also, the results of this research can be applied to identify valuable travellers on marketing or CRM systems of the airline industry, and the framework can be applied to other industries. Figure 1 shows that the growth rate of departing passengers in 2007 was lower than the previous rates. The passenger growth rates of departing passengers declined from 2007 to 2009 due to the high price of aviation fuel and global economic depression, the growth rates of arriving passengers were from 3,378,118 to 5,565,681 (from 2005 to 2010). The growth rates of arriving passengers were found to decline from 14.50 to 3.47% (from 2005 to 2008) and to increase up to 14.30% from 2008 to 2009. The growth rate of 2010 increased about 12.34% from 2009. In 2013 and 2014, both of the growth rates increased. Hence, the ratio of departing/arriving passengers in 2013 and 2014 was 1.3788 and 1.1952. That is, Taiwan government still had to attract international tourists to visit Taiwan for balancing the ratio. However, one of the reasons for increasing the growth rate is that the low-price airlines (such as Jet Airways, V Air, Air Asia and Tigerair Taiwan) get involved in Taiwan airline markets. As Figure 1 illustrates, the departing passengers increased from 8,208,125 to 13,182,976 from 2005 to 2015. The Tourism Bureau of Taiwan (2016) reported that most arriving passengers visited Taiwan for the purpose of tourism, the share of tourism market was about 42% (5) each year (from 2005 to 2015). Compared to 2004, the growth rate of arriving passenger in 2005 was only about half value; the rates from 2006 to 2008 were both lower than 6%. The main factors might be due to the global economic depression and high unit price of aviation fuel. Although, the growth rate of 2009 was 14.30%. Therefore, the Tourism Bureau of Taiwan is still having to implement a marketing project for arrival and departure of passengers of Taiwan. Hence, the airlines (China Airlines, Eva Airways, TransAsia Airways and Mandarin Airlines) and travel agencies in Taiwan have to implement database-marketing projects to their membership travellers as well.

2. Literature review

2.1 RFM model and customer value

The RFM model is widely used for a variety of retailing industries, where the R (Recency) is defined as Recent purchasing date; the F (Frequency) is defined as Frequency of purchasing in a specific period; and the M (Monetary) is defined as Average amount of purchasing in a specific period (Hughes 1994). For usage of RFM model, each variable of the RFM model can be divided into five equal portions. The first portion is the lowest level, and the fifth portion is the highest level. Thus the RFM model can be classified from 111 to 55 5 (Miglautsch 2000), and there are a total of 125 grades. Nevertheless, 125 grades are not clearer for observing, five levels of the RFM model can be reduced into two levels; that is, customer values can be clearly observed by eight grades (two levels: low and high, Shih and Liu 2003, Lin and Tang 2006; Chiang 2014). Customer value can be estimated via the RFM model, and the RFM model can be processed in market segmentation to find some valuable customers (Goodman 1992). For example, Lin and Tang (2006) applied the RFM model to analyse customers of music products. They classified homogeneous customers in the same group. Their study classified the RFM variables into 2 levels: High and Low. Thus, eight clusters were established. Also, in their research, the Apriori algorithm (Agrawal, Imielinski, and Swami 1993) was employed to create association rules. Wong et al. (2006) adopted the RFM model to estimate traveller values in Taiwan. Their research found seven segments via C4.5 decision tree, and they also discovered 12 association rules of travellers next preferred destinations by Apriori algorithm. Hence, their research results can be applied on marketing systems of airlines. Lumsden, Beldona, and Morrison (2008) also applied the RFM model on travel firms for predicting customer values of the target market in travel vacation clubs. Furthermore, Khajvand and Tarokh (2011) evaluated customer value based on weighted RFM model for making decisions in CRM systems of bank industry; Shim, Choi, and Suh (2012) also proposed VIP or non-VIP for the RFM model. They applied a new model and data mining for reaching customer rules of online stores. Moreover, in the field of classification and segmentation with RFM model, Cheng and Chen (2009) employed the RFM model and K-means algorithm into Rough set theory (Pawlak 1982) to mind accuracy classification rules which can be applied in a CRM system. Also in this field, Wei et al. (2013) applied Self Organising Map (SOM) and K-means algorithms on the RFM model for partitioning customers of hair salons in Taiwan. Their research found the market were of four types: loyal, potential, new, and lost customers. RFM model can be applied in specific industries. For instance, Chao and Yang (2003) applied Back-propagation Neural Network on the transactions of the medical equipment industry based on the RFM model for predicting customer values. Shih and Liu (2003) applied Analytic Hierarchy Process (AHP) to determinate the weights of the RFM model and sorted the customer values. This research found eight customer clusters and two types of high-value customers. Moreover, Wong and Chung (2007) applied the RFM model and the decision tree algorithm to analyse Taiwans domestic airline passengers, the results could identify valuable passengers, and it might help domestic airlines of Taiwan for ticket sales. For the researches of improvement of RFM model. In Chiangs study (2010), the RFM model has been improved to be RFMP (P: Price-Discount) for mining useful customer value of online shopping markets. The study found three markets of online shoppers: Cost, Risk, and Convenience; there were five and four association rules established for urban and suburban university students in Taiwan. Besides, for analysing customers of pharmaceutical industry in Taiwan, Chuang, Chia, and Wong (2013) employed C5.0 decision tree algorithm and FMC model (RFM model based, C for contract term) on classifying health care institutions in Taiwan. However, their research found four types of health care institutions for enhancing their marketing strategy. However, discovering customer values via the RFM model is not proper for any industry, businesses may develop a new model of customer values for their own applications.

2.2 Apriori algorithm and marketing system

Association rules are widely used in a variety of business and science managements. The most commonly used algorithm for mining association rules is the Apriori algorithm (Agrawal, Imielinski, and Swami 1993). The purpose of the Apriori algorithm is to scan the database for combinations of related information candidates (Itemset). Then, to calculate support value for each itemset. The itemset is for determining whether the combinations are classified in the database for association rules group. As for the research of online shopping behaviours, Yang and Lai (2006) applied basket market analysis on online shopping behaviours, their research data were collected from a POS (point of sales) system, their research results found significant association rules for decision-making of bundling sales. In 2003, Chen and Chang applied the RFM model to partition customers for obtaining higher value customers. Their study employed the Apriori algorithm to analyse relationships among pharmaceutical products for mining possible product portfolios of sales. The results found eight product portfolios that can increase shopping amount from 2.5 to 4.5 products. For enhancing CRM system, Gong et al. (2007) applied Apriori algorithm on a CRM system of a carpet company for reducing invalid records and scanning times. The CRM system was improved for saving the time efficiently. Association rules can also be applied in discovering patterns for developing new product. For example, Liao et al. (2011) employed Apriori algorithm and clustering method on marketing system of an international travel agency in Taiwan. Their research discovered customer knowledge patterns, and the patterns can be applied to developing new travel products for travel agencies. Since the researches regarding the improvement of the Apriori algorithm, Chiang (2010) applied the Apriori algorithm for supervised database. The objective of Chiangs paper was to discover behaviours of community courseselection. In Chiangs research, the fuzzy cluster method was applied to segment students, and applied the improved supervised-fuzzy-Apriori (SAA, Chiang 2008) for generating fuzzy association rules. The rules can be employed on marketing projects of university community curriculums. However, for retail industries, association rule is a useful data mining technology, which can help retailers to generate customer shopping-behaviour rules (Bilgic, Kantardzic, and Cakir 2015). These useful rules can be applied in marketing or CRM systems for implementing target marketing strategies efficiently.