Abstract
Everyday, millions of people travel around the globe for business, vacations, sightseeing, or other reasons. An astronomical amount of money is spent on tickets, accommodations, food, transportation, and entertainment. According to World Travel and Tourism Council, travel and tourism represents approximately 11% of the worldwide gross domestic product (GDP) (Werthner & Ricci, 2004). Tourism is an information-based business where there are two types of information flow. One flow of information is from the providers to the consumers or tourists. This is information about goods that tourists consume such as tickets, hotel rooms, entertainments, and so forth. The other flow of information which follows a reverse direction consists of aggregate information about tourists to service providers. In this chapter we will discuss the second form of information flow about the behavior of tourists. When the aggregated data about the tourists is presented in the right way, analyzed by the correct algorithm, and put into the right hands, it could be translated into meaningful information for making vital decisions by tourism service providers to boost revenue and profits. Data mining can be a very useful tool for analyzing tourism-related data.
TopBackground
According to Tan, Steinbach, and Kumar (2006), “Data mining is the process of automatically discovering useful information in large data repositories” (p. 2). It uses machine learning and statistical and visualization techniques to discover and present knowledge in a form that is easily comprehensible to humans. Data mining involves four key steps: (1) data collection, (2) data cleaning, (3) data analysis, and (4) interpretation and evaluation. During data collection the most suitable data need to be collected from the most appropriate sources. There is often a need to consolidate data from a number of sources. Cleaning or cleansing is the process of ensuring that all values in a data set are consistent and correctly recorded (Hui, Pandey, Steinbach, & Kumar, 2006). Obvious data errors are detected and corrected, and missing data is replaced in this step. The third and the most important stage of data mining is the analysis of the data using known techniques. Usually the analysis is done using statistical- or machine-learning-based approaches. The choice of the technique depends on the type of problem and also the availability of appropriate data mining software (Bose & Mahapatra, 2001). The most difficult part in data mining is interpretation of results obtained during data analysis. The analysis results may show a high degree of accuracy, but unless the accuracy can be related to the context of the data mining problem, it is of no use. Once some meaningful interpretation of the data analysis results can be done, the final step is to take action(s) so that the gathered knowledge can be put into practical use. In the case of tourism data mining, the data mining process goes through the same four steps. Figure 1 identifies the four steps involved in data mining and is similar to the one used in Bose and Pal (in press).
Figure 1. Different steps in data mining
Key Terms in this Chapter
Clustering: It is a technique that classifies instances into classes by calculating the distance between them and other instances. The instances that have the least distance between them are grouped into the same class. It is a type of unsupervised learning.
Support Vector Machines (SVM): SVM is a data mining method useful for classification problems. It uses training data and kernel functions to build a model that can appropriately predict the class of an unclassified observation.
Mean Absolute Percentage Error (mape): MAPE is used as a figure of merit to identify whether a data mining method is performing well or not. The lower the MAPE, the better the performance of the data mining method.
Genetic Algorithms: These algorithms mimic the process of natural evolution and perform explorative search. The main component of this method is chromosomes that represent solutions to the problem. It uses selection, crossover, and mutation to obtain chromosomes of highest quality.
Artificial Neural Networks (ANN): ANN is a pattern matching technique that uses training data to build a model and uses the model to predict unknown samples. It consists of input, output, and hidden nodes and connections between nodes. The weights of the connections are iteratively adjusted in order to get an accurate model.
Tourism Recommendation System: Information technology may be used to make recommendations to tourists about sightseeing venues, restaurants, entertainment locations, shopping, and so forth. These systems will revolutionize future travel by making recommendations using data on demographics and preferences of tourists.
Self-Organizing Feature Map (SOFM): SOFM is a data mining method used for unsupervised learning. The architecture consists of an input layer and an output layer. By adjusting the weights of the connections between input and output layer nodes, this method identifies clusters in the data.
Decision Tree: It is technique for classifying data. The root node of a decision tree represents all examples. If these examples belong to two or more classes, then the most discriminating attribute is selected and the set is split into multiple classes.
Rough Sets (RS): RS is a technique used for classification. It is based on the notion of approximate class membership of an unclassified observation. The main steps in RS analysis include discretization, formation of reducts, and enumeration of rules.