Research Article
Visitors' Spending on Accommodation: A Segmentation Model using Two-Step
CHAID Analysis
Rosmini Ismail1*, Hartini Jaafar2, Ramlee Ismail3, Khalizul Khalid4 1*,2,3,4Faculty of Management and Economics,
Universiti Pendidikan Sultan Idris, Tanjong Malim, Perak, Malaysia rosmini@fpe.upsi.edu.my1*
Article History: Received: 10 November 2020; Revised: 12 January 2021; Accepted: 27 January 2021;
Published online: 05 April 2021
Abstract: Tourism receipt through visitors' spending is one of the contributors that stimulate the local economy. Therefore, it
is crucial to determine the factors influencing these spending preferences. This study determines factors and average spending on accommodations using segmentation techniques for Perhentian Island's visitors. Determinant factors include demographic, trip-related, and psychographic characteristics. Data were collected through a survey and run for 929 visitors using two-step Chi-Square Automatic Interaction Detection (CHAID) analysis. The analysis produces a three-level regression tree and later a classification tree. The findings documented that, level 1 consisted of four segments and were segmented according to the country of origin (COO). Overall, the Italian is a segment that has the highest average spending. The fourth segment of level 1, namely Malaysia, branched out further to level 2 and level 3. These levels were segmented based on the number of dependents during the trip and length of stay, respectively. For domestic visitors, Malaysian with dependents on the trip spend the highest. Based on the results, recommendations for the Perhentian Islands accommodation operator were to provide infrastructure to accommodate families for domestic market and marketing strategy that target Italians for the international market. The results could also assist local authorities outlining tourism planning.
Keywords: Visitors spending, accommodations, CHAID, segmentation, Perhentian Island
1. Introduction
The growing needs of modern societies for recreational and leisure have turned tourism activities into a complex industry. As a result, the tourism industry is growing, and it is reported that business volumes within the sectors are equals to, if not surpassing the oil export, food products or automobiles (World Tourism Organisation (UNWTO), 2016). Simultaneously, reports from the World Tourism Organization (UNWTO) documented that the year 2012 marked as the first time for attaining more than 1 billion international tourist arrivals worldwide (World Tourism Organization (UNWTO), 2013). During this period, Malaysia is in the top ten destination arrivals and afterwards, continue to show increasing tourism receipt with reported RM 69.1 billion (USD 17.27 billion) in 2015, RM 82.1 billion (USD 20.52 billion) for 2016, 82.2 billion (USD 20.55 billion) in 2017 and 84.1billion (USD 21.02 billion) for 2018 (Tourism Malaysia, 2019). It is an indicator that visitors spending is the driving force of the tourism industry. Due to this, some countries require visitors to bring in a certain amount of foreign currency for each day of their stay (Welgamage, 2015).
Tourists play a vital role in the tourism industry because their expenditures generate income that stimulate the growth of other sectors (Ghanem, 2017). Therefore, it is critical to understand their activities, experiences and spending (McCabe, 2009; Pratt, 2015). One of the suggestion is through tourist profiling studies using market segmentation approach (Abbruzzo, Brida, & Scuderi, 2014; Díaz-Pérez & Bethencourt-Cejas, 2016; Pratt, 2015). Through this approach, tourists were grouped into segments according to demographic, trip-related characteristics and other variables to predict their behaviour and preferences (Dolnicar, Grün, & Leisch, 2018; Mortazavi & Lundberg, 2019). There are various analysis techniques proposed, and one of them is the Chi-squared Automatic Interaction Detector (CHAID).
As the name suggests, the technique automatically detect association within data as opposed to conducting individual tests manually through inferential statistics tests such as differences in means and correlations. Even though there were various procedures available, CHAID tree-based were commonly utilised for tourists segmentation. CHAID was considered as a multidimensional method that proven to analyse better than discriminant analysis (Díaz-Pérez & Bethencourt-Cejas, 2016). Therefore, this study proposed to segment spending on accommodation for visitors of Perhentian Island, Malaysia according to demographic, trip-related and psychographic factors by utilising CHAID procedure.
2. Literature Review
Tourism studies on tourists evolved from defining tourists as a single group to homogeneous sub-groups through segmentation approaches (Sharma & Kumar Nayak, 2019). The purpose of segmentation studies is to define homogeneous groups of consumer within the scope of general consumer characteristics and situation-specific or behavioural characteristics (Legohérel & Wong, 2006). As a result, segmentation techniques may take various perspectives.
As an example, Guttentag, Smith, Potwarka, and Havitz (2018) profiled 923 tourists that used the non-traditional accommodation arrangement, Airbnb, into five distinct motivation-based segments, namely Money Savers, Home Seekers, Collaborative Consumers, Pragmatic Novelty Seekers and Interactive Novelty Seekers. Meanwhile, a study by Rid, Ezeuduji, and Pröbstl-Haider (2014)) grouped 430 tourists for rural tourism in the Gambia into four distinct market segments namely heritage & nature, multi-experiences, multi-experiences & beach, and sun & beach seekers. Wen, Meng, Ying, and Belhassen (2020) on the other hand, clustered 654 Chinese tourists based on their cannabis consumption during vacation and categorised them as cannabis enthusiasts, recreationists and the curious.
Apart from motivation-based, these types of studies were also concern with expenditure-based segmentation. It has been most intensively analysed in the recent decade (Mudarra-Fernández, Carrillo-Hidalgo, & Pulido-Fernández, 2019). These type of expenditure pattern modelling not only able to assists businesses, but for decision making by relevant authorities (Pulido-Fernández, Carrillo-Hidalgo, & Mudarra-Fernández, 2019). It allows for businesses and tourism development authority to predict and customise their offer to fulfil their target market segment's needs (Boztug, Babakhani, Laesser, & Dolnicar, 2015). Additionally, it leads to critical new insights and perspectives (Dolnicar, Grün, & Leisch, 2018).
In general, those who are vacationing in the same area at the same time may spend differently according to segments that could be characterised by one or several variable(s) (Legohérel & Wong, 2006). For example, Marrocu, Paci, and Zara (2015) categorised tourists spending into three segments, namely light, medium and heavy spender for non-resident tourists who spent their holidays in Sardinia. It was found that for heavy spenders, they were significantly characterised by income, foreign nationality, previous visits and notoriety-motivated holidays. As for light-spending tourists, trip characteristics, especially party size and the number of visited sites contribute to reducing the level of tourism expenditure.
Meanwhile, Saayman and Saayman (2018) uses visitor spending to determine scuba divers expenditure surrounding the Portofino Marine Protected Areas (MPA). On average, scuba divers spending were €400 per trip or €110 per day. The average spending can be clustered according to six segments which are local rescuer divers, international big spender diver, intracontinental divemaster, new local diver, intracontinental advanced divers and local instructors. The study documented that low and medium spender consists of local divers while big spender was international divers.
As for Serra, Borges and Marujo (2016), their study documented significant differences in tourists spending from different countries in City of Evora in 2015. Through CHAID, tourists' expenditure can be characterised according to country of origin and segmented into two groups. Segment 1, which was represented by the USA, Germany, France, the United Kingdom, Belgium, Canada, The Netherlands and other countries from Europe, 43.7% of them spent above €141.50. Meanwhile, the rest of the countries spent between €71 and €141.50. The second segment, on the other hand, comprised of Portugal, Brazil, Spain and other countries of the world, where 65% of those spend the night, spent more than €71. For daily visitors, almost 50% spent between €21 and €70.
A study conducted by Amir, Osman, Bachok, and Ibrahim (2015; 2016) which also utilised CHAID, examined tourist expenditure pattern and documented that international tourists to Melaka City tend to spend more on accommodation and food and beverages compared to domestic tourists. The study suggested that it is due to the longer duration length of stay and preference for higher star rated hotels accommodation. Consequently, it was found that tourists to Melaka City spend heavily on accommodation with RM690 and below. This expenditure is represented by more than 90% of overall tourist where 48% were domestic tourists and 42% international tourists
Most of expenditure-based segmentation studies utilised individual microdata collected via survey. This type of data for segmentation studies has been found as more valuable than aggregated expenditure model (Perić, Dragičević, & Škorić, 2019). Analysis techniques, however differed between segmentation studies. There were a variety of methods utilised such as ARIMA (Koc & Altinay, 2007), Cluster analysis (Menor-Campos, Fuentes
Jiménez, Romero-Montoya, & López-Guzmán, 2020; Nurjannah et al., 2019; Ryu, Choi, & Cho, 2020), Finite mixture modelling (FMM) (Mortazavi & Lundberg, 2019; Pani, Sahu, & Majumdar, 2020) and the Chi-squared Automatic Interaction Detector (CHAID) (Amir, Osman, Bachok, & Ibrahim, 2015; Borges & Marujo, 2016; Diaz-Perez & Bethencourt-Cejas, 2017; Ismail, Jaafar, & Khalid, 2016; Ismail & Khalid, 2016; Kelley et al., 2019).
As the name suggests, CHAID is an application that automatically detects association within data to produce segments or nodes when there were significant differences between groups at p<0.05. It generates tree diagrams that enable the identification of variables that have the strongest interactions (Milanović & Stamenković, 2016). CHAID was considered more complex than multivariate techniques and proven to analyse better than discriminant analysis for these types of studies (Díaz-Pérez & Bethencourt-Cejas, 2016). Furthermore, CHAID has an advantage over other decision tree analysis, namely CART and Quest in terms of generating more rules of different complexity degrees (Lin & Fan, 2019). Therefore, this study determines segmented average spending on accommodations for Perhentian Island visitors in Malaysia segmented according to demographic, trip-related and psychographic factors through two-step CHAID analysis.
3. Methodology
The designated tourism site for the study is the Perhentian Islands which comprises of two islands called, Perhentian Kecil and Perhentian Besar. The main attractions for the islands are white sandy beaches and underwater scenery which include a spectacular view of coral reefs. The study employs a survey method by distributing questionnaires to collect demographic, trip-related, psychographic characteristics and visitors spending data. The questionnaire was developed by adapting and adopting items from past instruments and later validated by experts. A pilot study was conducted, and face validity was also tested. Estimated time allocated to fill in the questionnaire were between 25 – 30 minutes per respondents.
The primary study's data collection started with more than 1000 questionnaires distributed within eight months during the year 2016. Two-step decision tree analyses were conducted by employing the Chi-Square Automatic Interaction Detector (CHAID) procedure in SPSS 22.0 software. The procedure of analysis for CHAID involved splitting the test samples into various nodes by selecting the largest significant relationship between predictor and dependent variables. The resulting nodes were further divided into various nodes with smaller sample size by other descriptors. The splitting stopped if there was no significant difference between the variables.
The first step involved running a dependent variable represented by continuous scale spending data against predictor variables consisting of demographic, trip-related and psychographic factors to generate a regression tree. The outputs from regression were then, run again in CHAID to produce a classification tree model. In this study, the classification tree was pruned to increase the accuracy of the percentage correct classification to 100%. There were no rules of thumb regarding the correct classification percentage; the 100% set in the study was discretionary part of the researchers.
4. Findings and Discussions
The study gathered responses from 527 (57.4%) international tourists and the remaining 402 (42.59%) were domestics. International respondents for this study were from five continents namely Europe (75%); Asia (14.2%); 5.5% the Americas (North and South America), Oceania (4.2%) and the remaining 1.1% were from the Africa continent. The biggest proportion of continent representation of the study were European whilethe Asia continent made up of 14.2% from overall. The top five arrivals for countries were France, Germany, Italy, China and the United Kingdom, which made up more than 65% from overall respondents.
Background analyses were described based on the eleven predictor variables namely gender, age, income, marital status, types of tourists (international versus domestic), nationality (country of origin), location of the Island, companionship, motivation, length of stay and number of dependent for the trip. Other than that, psychographic information which includes value for money, overall experience, island crowdedness, and sufficiency of utilities provided, satisfaction of accommodation and other tourism services on the Island.
Descriptive Analysis
The demographic characteristics of the 929 respondents were illustrated in Table 1. Comparatively, an accumulated of 50% international respondents came from a broader range of age, 21 to 30, than domestic visitors
where half of them clustering the range 21 to 25 years of age. Meanwhile, gender groups represent quite a balance proportion for both international and domestics with 41% and 49.3% for males respectively, and 52.9 and 50.7% for females. Marital status, on the other hand, showed a considerable difference between international and domestic, where more than 70% of singles were domestics in comparison with 40.6% internationals.
As for monthly household income, it has been expected that there will be a massive gap in income range between international and domestics respondents which due to currency exchange values. As illustrated in Table 3, 78.3% domestics were mostly flocked to the income range of less than RM7,000(USD1750).
Meanwhile international mostly (60.5%) positioned between RM7,001(USD1750) to RM15,000.00(USD3750,00). It can be argued that domestics' income range corresponds to age range characteristics, where 59.4% were below 25 years of age which may be represented by students and fresh graduates who were just started working.
Table 1.Demographic characteristics
Predictor Variables
Detail descriptions International (%) Domestic (%) Age Below 21 4.4 8.7 21 – 25 26.4 50.7 26 – 30 34.9 19.7 31 – 35 15.7 12.2 36 – 40 7.2 6.0 41 – 45 6.5 1.7 Above 45 4.4 0.9 Missing value 6.1 - Gender Male 41.0 49.3 Female 52.9 50.7 Missing value 6.1 -
Marital Status Single 40.6 72.9
Married 19.9 21.6 Unmarried couple 38.3 3.5 Missing value 1.1 2.0 Monthly Household Income Less than RM3000 1.1 52.7 RM3001 – RM7000 7.0 25.6 RM7001 – RM11000 34.3 3.7 RM11001 – RM15000 26.2 4.2 RM15001 – RM19000 7.0 0.7 RM19001 – RM 23000 4.7 0.2 RM 23000 – RM27000 1.5 1.2 Above RM27000 6.3 0 Missing value 11.8 11.4 Number of dependent on this trip 0 41.7 76.4 1 51.8 10.5 2 1.9 6.7 3 3.2 2.2 4 1.1 1.0 5 0 2.2 6 .2 0.7 More than 6 0 0.2 Total respondents 527 402
Results for trip-related characteristics of respondents were illustrated in Table 2. International tourists located in Perhentian Kecil were 58.6% and 41.4% in Perhentian Besar. Domestic tourists, on the other hand, the majority (57.7%), stayed on Pulau Besar and the remaining 42.3%, on Perhentian Kecil. The motivation for visiting the Perhentian Island variable, which applies only to international tourists indicates that only 16.5% came to Malaysia solely for visiting Perhentian Island. Another 45.4% came to Malaysia were motivated to visit not only Perhentian but other places in Malaysia as well. Meanwhile, the remaining 28.8% were unaware of Perhentian Island, and the remaining 8.5% were visiting the Islands because it was included in the tour package. Overall, international tourists' average stays were four nights, and domestic's two nights stay.
Table 2.Trip Characteristics
Predictor Variables
Detail descriptions Int' l
(%)
Domestic (%)
Location Perhentian Kecil 58.8 42.3
Perhentian Besar 41.2 57.7
Companion- ship By myself 8.0 5.7
Family 17.1 13.7
Partner 41.0 8.0
Friends 24.5 65.4
Groups 9.5 7.2
Motivation for Visiting the Island
Yes, it is the only reason I'm coming to Malaysia 16.5 NA Yes, but I have other places to visit as well 45.4 NA No, I was visiting other places in Malaysia when
someone told me about Perhentian Island
28.8 NA
No, I have other places which are more interesting to visit. Perhentian island just happens to be in the package 8.5 NA Missing values .8 NA Length of Stay 1 .2 15.7 2 3.4 56.0 3 17.5 15.9 4 23.5 8.0 5 24.5 4.0 6 12.1 - 7 6.5 - 8 days 1.9 - 9 days 1.1 0.5 10 days 5.7 -
More than 10 days 3.6 -
Total respondents 527 402
Psychographic characteristics point to the opinion of tourists in regards to their trip to Perhentian Island's worthiness, island crowdedness and satisfaction. The first and second items measured whether respondents felt that the money they spent is worthy of the experience gained and the extent of satisfactory experience during the trip.
The result for the items indicates favourable opinion with mean 7.21 and 7.76, respectively. Meanwhile, the third and fourth items evaluate perceived crowdedness by respondents, which include Island crowdedness as a whole and accommodation crowdedness. For this, the means were 5.36 and 4.46, indicating neutral or 'on the fence' views regarding the matter. The remaining of psychographic characteristics measured satisfactory towards accommodation and other tourism services on the Island. Both items show means within the range of 5.5 and 6.0. The results can be interpreted as slightly satisfied with accommodation and other tourism services on the Island.
Chi-squared Automatic Interaction Detector (CHAID) Models
A regression tree was generated to obtain segmented average spending based on demographic, trip-related and psychographic factors. Through CHAID, determinant factors were automatically detected, and nodes were produced, consisting of average spending values. These node (s) were further split or branched out when there were significant differences between groups at p<0.05. Nodes that do not split further or branched out were designated as segments. As illustrated in Figure 1, the regression tree for spending on accommodation consisted of three levels, segmented by six predictor variables which generate sixteen nodes, denoted by Node 1 to Node 16.
The overall regression tree showed that level 1 has four nodes with Node 4 designated as Segment 1 and level two has six nodes with three nodes designated as Segment 2, Segment 3 and Segment 4.
Meanwhile, level 3 has six nodes, and all of the nodes were designated as segments (Segment 5 to Segment 10). Therefore, there were ten values of segmented average spending on accommodation as depicted by (i) Node 2; (ii) Node 7; (iii) Node 8; (iv) Node 10; (v) Node 11; (vi) Node 12; (vii) Node 13; (viii) Node 14; (ix) Node 15 and, (x) Node 16. Next, Figure 1a and Figure 1b were reproduced from Figure 1 as the enlarged version of the regression tree.
Figure 1. Regression Tree for Spending on Accommodation
Figure 1a demonstrated levels 1 and 2 of the regression tree. At level 1, the predictor variable, namely country of origin, was split into four nodes. The first node, Node 1, comprised of six countries/group of countries, namely (i) France, (ii)Germany, (iii) Spain, (iv)other European countries, (v)Other Asian countries and (vi) the USA and Canada. Meanwhile, Italy independently fell under Node 2, followed by another six countries (i) the United Kingdom, (ii) the Netherlands, (iii) Sweden, (iv) China, (v) Africa, and (vi) Australasia in Node 3.
Finally, the last node of level 1 represented Malaysia denoted by Node 4. Since Node 2, which comprised by Italy, do not split further into level 2, it was designated as Segment 1 with segmented average spending of RM741 (USD185). It is important to emphasise that the CHAID procedure automatically clusters the countries comprised within a node. Next, the second level of the regression tree was segmented according to two predictor variables, namely monthly household income and number of dependents. Three nodes do not split further to level 3 and therefore designated as segments. Segment 2 represented those from the UK, Netherlands, Sweden, China, Africa, Australia and New Zealand who visited the Island without dependent, incurring average spending on accommodation of RM351 (USD88).
Meanwhile, Segment 3 were those from the same clusters of countries but with dependent on the trip. Their spending was on average RM528 (USD132) for accommodations. Segment 4, on the other hand, representing Malaysian with dependent to the Island, spending an average of RM215 (USD54).
Figure 1a. Regression Tree Level 1 and 2 for Spending on Accommodation (reproduce from Figure 1)
The regression tree was further branched out to the third level as illustrated and enlarged in Figure 1b. At this level, there were three predictor variables, namely companionship, gender and length of stay. Each variable produced two nodes, which brings the total of six nodes at the third level. All of the nodes at this levels were designated as segments. All four segments (Segment 5, Segment 6, Segment 7 and Segment 8) comprised of those from the same cluster of countries namely France, Germany, Spain, Other Europe, Other Asia, USA, Canada and South America. However, Segment 5 and Segment 6 were those with a monthly household income ranging from RM7000 - RM11000 (USD1750- USD2750) or less, whereas, Segment 7 and Segment 8 earned above that range. Segment 6, Segment 7 and Segment 8 came from the same clusters of countries. Node 11 until Node 16, was at the third level of the tree, which means, for these segments, they were segmented according to these three variables.
Next, classification trees were generated to estimate the proportion for the average spending and to determine the model's predictive classification power. Similar to the regression tree, each level of the classification tree may be segmented by one or more predictor variables. However, the further the levels a tree produced, the lower the predicted correct classification percentage may be. Therefore, the classification tree generated by CHAID for this study was set at a maximum of three levels only. Afterwards, the classification tree was pruned until the predictive power of 100% correct to achieve the highest accuracy for the model. However, this process may cause several variables were being filtered out. For example, transforming the regression tree (Figure 1) into a classification tree (Figure 2) has resulted in the trees were not split at the second level for countries other than Malaysia as in their regression trees counterpart.
Figure 1b. Regression Tree Level 2 and 3 for Spending on Accommodation (reproduce from Figure 1)
Therefore, rather than having ten average spending values in the regression tree, the classification tree for accommodation has only six average values (refer to Table 3).
Overall, international tourists are illustrating higher spending on accommodation, and this aligned with other research which was conducted in Melaka City. Amir et al. (2015) found that international tourists in Melaka City tend to stay longer and prefer more luxurious hotels, thus contributing to higher spending on accommodation than the other four expenditure categories. In several studies, the type of accommodation was found to be one of the determinants of travel expenditure (Abbruzzo, Brida, & Scuderi, 2014; Guttentag et al., 2018). It is important to note that none of the psychographic factors were significant to influence visitor spending.
Table 3.Regression and Classification Tree Average Spending Comparison
Regression tree model (Predicted classification power = 83.2% Classification tree model; predicted power = 100% (i)France; (ii)Germany,
(iii)Spain, (iv)Other European countries, (v)Other Asian countries, (vi)the USA and Canada
Income less than
RM11000
Visit island with
other than friends RM 286
RM285 Visiting Island with
friends RM159 Income more than RM11000 Female RM338 Male RM492 (i)Italy RM741 RM741
(i)the United Kingdom, (ii)the Netherlands, (iii)Sweden, (iv)China, (v)Africa, and
(vi)Australasia
One or more dependent on the trip RM351
RM351 Zero dependent on the trip RM52
(i)Malaysia
One or more dependent on the trip RM215 RM215 Zero
dependent on the trip
Staying two nights or
less RM87 RM87
Staying more than
two nights RM 161 RM 161
5. Conclusions
The notion of the study is to determine factors that influence and segmented average spending on accommodation for Perhentian Island visitors. The results provide input to local authority and businesses for tourism planning and developing marketing strategy. The study found that their spending can be segmented according to a demographic factor, namely country of origin variable. Average spending of those from Italy is the highest from any other country with RM741 (USD185) per visit. International visitors can be segmented by this variable only.
However, Malaysian visitors' spending can be further segmented according to two trip-related characteristics, namely the number of dependent on the trip and length of stay. It was found that those with no dependent on the trip and stay two nights or less spend on average of RM87 (USD22).Meanwhile, those who stay more than two nights spend RM161 (USD40) per visit. As for Malaysian with dependents on the trip, they spend an average of RM215 (USD54) per visit. It is hoped that the findings of the study can accommodate local authorities and accommodation operators in formulating planning and strategy.
References
1. Abbruzzo, A., Brida, J. G., & Scuderi, R. (2014). Determinants of individual tourist expenditure as a network: Empirical findings from Uruguay. Tourism Management, 43(0), 36-45. doi: http://dx.doi.org/10.1016/j.tourman.2014.01.014
2. Amir, S., Osman, M. M., Bachok, S., & Ibrahim, M. (2015). Understanding domestic and international tourists' expenditure pattern in Melaka, Malaysia: Result of CHAID analysis. Procedia-Social and Behavioral Sciences, 172, 390-397.
3. Amir, S., Osman, M. M., Bachok, S., & Ibrahim, M. (2016). Local Economic Benefit in Shopping and Transportation: A study on tourists' expenditure in Melaka, Malaysia. Procedia-Social and Behavioral Sciences, 222, 374-381.
4. Boztug, Y., Babakhani, N., Laesser, C., & Dolnicar, S. (2015). The hybrid tourist. Annals of Tourism Research, 54, 190-203.
5. Díaz-Pérez, F. M., & Bethencourt-Cejas, M. (2016). CHAID algorithm as an appropriate analytical method for tourism market segmentation. Journal of Destination Marketing & Management.
6. Diaz-Perez, F. M., & Bethencourt-Cejas, M. (2017). An Application of the CHAID Algorithm to Study the Environmental Impact of Visitors to the Teide National Park in Tenerife, Spain. International Business Research, 10(7), 168-177.
7. Dolnicar, S., Grün, B., & Leisch, F. (2018). Market Segmentation Market Segmentation Analysis: Understanding It, Doing It, and Making It Useful (pp. 3-9). Singapore: Springer Singapore.
8. Mr. Dharmesh Dhabliya, Ms. Ritika Dhabalia. (2014). Object Detection and Sorting using IoT. International Journal of New Practices in Management and Engineering, 3(04), 01 - 04
9. Ghanem, J. (2017). Conceptualising "the Tourist": A critical review of UNWTO definition.
10. Guttentag, D., Smith, S., Potwarka, L., & Havitz, M. (2018). Why tourists choose Airbnb: A motivation-based segmentation study. Journal of Travel Research, 57(3), 342-359.
11. Ismail, R., & Khalid, K. (2016). Profiling Tourists According to Spending Behaviour: Examining Perhentian and Pangkor Islands Visitors. In A. M. Abdullah, K. W. Yahya, N. Ramli, R. S. Mohamed & E. B. Ahmad (Eds.), Regional Conference on Science, Technology and Social Sciences (RCSTSS 2014): Business and Social Sciences (pp. 1041-1052). Singapore: Springer Singapore.
12. Ismail, R., Jaafar, H., & Khalid, K. (2016). Estimating Perhentian Island's Environmentally Sustainable Tourism Yield for the Recreational Segment. Asia Pacific Journal of Innovation in Hospitality and Tourism, 5(3), 51 - 70.
13. Kelley, K. M., Bruwer, J., Zelinskie, J., Gardner, D. M., Govindasamy, R., Hyde, J., & Rickard, B. J. (2019). Travel group member type effects in wine tourism: an ECHAID segmentation. Tourism Recreation Research, 44(1), 54-65.
14. Koc, E., & Altinay, G. (2007). An analysis of seasonality in monthly per person tourist spending in Turkish inbound tourism from a market segmentation perspective. Tourism Management, 28(1), 227-237. doi: http://dx.doi.org/10.1016/j.tourman.2006.01.003
15. Legohérel, P., & Wong, K. K. (2006). Market segmentation in the tourism industry and consumers' spending: What about direct expenditures? Journal of Travel & Tourism Marketing, 20(2), 15-30. 16. Lin, C.-L., & Fan, C.-L. (2019). Evaluation of CART, CHAID, and QUEST algorithms: a case study of
construction defects in Taiwan. Journal of Asian Architecture and Building Engineering, 18(6), 539-553.
17. Marrocu, E., Paci, R., & Zara, A. (2015). Micro-economic determinants of tourist expenditure: A quantile regression approach. Tourism Management, 50(0), 13-30. doi: http://dx.doi.org/10.1016/j.tourman.2015.01.006
18. McCabe, S. (2009). Who is Tourist? Conceptual and Theoretical Developments. In J. Tribe (Ed.), Philosophical Issues in Tourism. Bristol: Channel View Publications.
19. Menor-Campos, A., Fuentes Jiménez, P. A., Romero-Montoya, M. E., & López-Guzmán, T. (2020). Segmentation and sociodemographic profile of heritage tourist. Tourism and Hospitality Management, 26(1), 115-132.
20. Milanović, M., & Stamenković, M. (2016). Chaid decision tree: methodological frame and application. Economic Themes, 54(4), 563-586.
21. Mortazavi, R., & Lundberg, M. (2019). Expenditure-based segmentation of tourists taking into account unobserved heterogeneity: The case of Venice. Tourism Economics, 1354816619841713.
22. Mudarra-Fernández, A. B., Carrillo-Hidalgo, I., & Pulido-Fernández, J. I. (2019). Factors influencing tourist expenditure by tourism typologies: a systematic review. Anatolia, 30(1), 18-34.
23. Nurjannah, Astuti, A. B., Fernandes, A. A. R., Amaliana, L., Yanti, I., & Isaskar, R. (2019). Two Step Cluster Analysis for Tourist Segmentation Coastal Object for Green Marketing Strategy. Paper presented at the IOP Conference Series: Earth and Environmental Science.
24. Pani, A., Sahu, P. K., & Majumdar, B. B. (2020). Expenditure-based segmentation of freight travel markets: Identifying the determinants of freight transport expenditure for developing marketing strategies. Research in Transportation Business & Management, 100437.
25. Perić, M., Dragičević, D., & Škorić, S. (2019). Determinants of active sport event tourists' expenditure– the case of mountain bikers and trail runners. Journal of Sport & Tourism, 23(1), 19-39.
26. Pratt, S. (2015). The economic impact of tourism in SIDS. Annals of Tourism Research, 52, 148-160. 27. Pulido-Fernández, J. I., Carrillo-Hidalgo, I., & Mudarra-Fernández, A. B. (2019). Factors Influencing
Tourism Expenditure on Accommodation in World Heritage Cities Trends in Tourist Behavior (pp. 209-224): Springer.
28. Rid, W., Ezeuduji, I. O., & Pröbstl-Haider, U. (2014). Segmentation by motivation for rural tourism activities in The Gambia. Tourism Management, 40(0), 102-116. doi: http://dx.doi.org/10.1016/j.tourman.2013.05.006
29. Ryu, S., Choi, K., & Cho, D. (2020). A behaviour-based typology of travellers using an online travel marketplace. Current Issues in Tourism, 1-19.
30. Saayman, M., & Saayman, A. (2018). Are there economic benefits from marine protected areas? An analysis of scuba diver expenditure. European Journal of Tourism Research, 19, 23-39.
31. Serra, J., Borges, M., & Marujo, N. (2016). Expenditure pattern analysis of cultural tourists at World Heritage Sites: an empirical evidence based on CHAID analysis. In Tourism, Lifestyles and Locations: Book of Extended Abstracts of ATLAS Annual Conference (pp. 96-102).
32. Sharma, P., & Kumar Nayak, J. (2019). An analysis on the emotional approach to segmentation: A study of yoga tourism. Paper presented at the Journal of Convention & Event Tourism.
33. Tourism Malaysia. (2019). Malaysia Tourism Statistics in Brief. Retrieved 1st February 2019, from http://www.tourism.gov.my/statistics
34. Welgamage, P.-P. L. (2015). Tourism economics in Sri Lanka: An econometric analysis. International Journal of Business and Social Research, 5(1), 90-101.
35. Wen, J., Meng, F., Ying, T., & Belhassen, Y. (2020). A study of segmentation of cannabis-oriented tourists from China based on motivation. Current Issues in Tourism, 23(1), 36-51.
36. World Tourism Organisation (UNWTO). (2016). Tourism - an economic and social phenomenon. Retrieved 1st February 2016, from http://www2.unwto.org/content/why-tourism
37. World Tourism Organization (UNWTO). (2013). UNWTO Tourism Highlights 2013 Editions. In UNWTO (Series Ed.) (2013 ed.). Madrid, Spain: UNWTO.
