تقسیم بندی بازار : کاربرد شبکه های عصبی

The objective of the research is to consider a self-organizing neural network for segmenting the international tourist market to Cape Town, South Africa. A backpropagation neural network is used to complement the segmentation by generating additional knowledge based on input–output relationship and sensitivity analyses. The findings of the self-organizing neural network indicate three clusters, which are visually confirmed by developing a comparative model based on the test data set. The research also demonstrated that Cape Metropolitan Tourism could deploy the neural network models and track the changing behavior of tourists within and between segments. Marketing implications for the Cape are also highlighted.

Marketing an international tourism destination such as Cape Town in South Africa has never been more dynamic, competitive, and important than it is today. Successful marketing requires careful planning and comprehensive analysis of data and information obtained from tourists that frequent destinations and those that do not. There is no shortcut to establishing a positioning strategy that could deliver a valuable experience to tourists. The ability to identify and serve tourists and create a dialogue with them has become a necessity for destination organizations such as Cape Metropolitan Tourism (hereafter referred to as Cape Metro Tourism). Managing the tourist relationship has become an essential part of attracting those with specific profiles to a destination. All activities around the tourist “touch points”––which aim to identify, attract, and retain the most valuable tourists for a destination and its enterprises––should be considered. The end result is to enhance retention and loyalty and sustain growth from profitable tourists. It is important to determine what it takes to encourage them to purchase the product/service that a destination offers. A need exists to understand their behavior and thus more in-depth knowledge about the homogeneous characteristics of groups is required over and above evaluating overall arrivals, expenditure figures and trip characteristics. Marketing strategists of destination organizations often encounter the problem of how to appropriately segment the market and package differentiated products and services for target segments. Segmentation is a methodological process of dividing a market into distinct groups that might require separate experiences or marketing service mixes (Venugopal and Baets 1994). Customer clustering is one of the most important techniques used to identify these segments (Saarenvirta 1998). Various clustering techniques are used as part of a methodology to identify segments, which become the foci of marketing strategy. The basis of this generally includes the identification and assessment of various tourist characteristics (such as demographics, socioeconomic factors, and geographic location) and product related behavioral characteristics (such as purchase behavior, consumption behavior, and attitudes towards and preference for attractions, experiences and services). Target marketing is a strategy that aims at grouping a destination’s markets into segments so as to aim at one or more of these by developing products and marketing programs tailored to each (Kotler 2001). Inadequate segmentation and clustering problems could cause a tourism destination organization, such as Cape Metro Tourism, to either miss a strategic marketing opportunity or not cash-in on the rewards of a tactical campaign. Market segmentation has developed as a methodology to identify target segments, with the outcomes of the process used to help understand tourists’ relationship with the destination. The objective of the research is to consider the use of a self-organizing (SOM) neural network for segmenting the international tourist market to Cape Town. A backpropagation (BP) neural network (based on the output provided by the former) was also used to complement the process by generating additional market knowledge about the relationship between the inputs used and the macrosegments obtained from the application of the SOM model. Input–output relationship and sensitivity analyses were used for the purpose of extracting additional market knowledge about the macrosegments.

The research findings suggest that the methodology and application of a SOM neural network is useful for segmenting the international tourist market of Cape Town. Furthermore, using a BP neural network model also provides additional market knowledge and, together with the SOM approach for segmenting the tourist market, may improve Cape Metro Tourism’s understanding of the international market. Currently, the organization only conducts descriptive analysis of the research data collated from international tourists and undertakes no modeling of the nature and scope described in this paper. Due to the predictive nature of neural networks, Cape Metro Tourism has an opportunity to use the surveys conducted on a continuous basis to track and understand changes in the behavior and profiles of tourists within and between the macrosegments over time. The deployment of the neural network models appears to have merit, even if the intention is only to use the technology as a basis to enhance the market strategists’ and media planners’ understanding of changing behavior among tourists within and between the macrosegments. If this were the rationale for deployment of the neural networks, various microsegments could be identified with an opportunity to disaggregate data and develop macrosegments. The use of the segmentation model will assist to achieve these analytical objectives. The introduction of alternative assessment tools––such as predictive techniques and the ability to assess patterns through the interaction of variables and not independently––also offers Cape Metro Tourism an opportunity to expand their research horizon: by focusing on the discovery of new profitable tourist segments and by identifying international tourists who do not visit Cape Town, but have a similar profile of the tourists frequenting