An Empirical Study of Passengers’ Perceived Satisfaction with Monorail Service Quality: Case of Kuala Lumpur, Malaysia

Abstract

The aim of the current study was to examine passengers’ perceived satisfaction with a monorail system by determining the factors that dominated their perception. For this study, 417 data points were collected through face-to-face questionnaire surveys in Kuala Lumpur, Malaysia, from 20 September 2019 to 10 December 2019. The study involved the use of several assessments to ascertain how the perceived satisfaction of passengers with the monorail service was influenced. The tests predicted which service factors fuelled this satisfaction and included an exploratory factor analysis, Spearman’s correlation test and an artificial neural network (ANN) model (termed the multilayer perceptron neural networks model) with a feed-forward backpropagation algorithm. The findings that were produced by these methods of analysing factors revealed the extraction of eight service-quality features as the key influences on the perceived satisfaction of passengers. The correlation test results revealed that these factors have a significant and positive relationship with perceived satisfaction. Finally, the ANN model with the optimum neuron number in the hidden layer is seven neurons. This model found that the dominant service quality factors that contributed to influencing passengers’ perceived satisfaction levels were (i) the provision of information, (ii) facilities, (iii) signage. The results of this study will benefit service providers, policymakers and planners in formulating effective strategies to enhance passengers’ satisfaction with the monorail service and increase the ridership.

1. Introduction

A crucial aspect of the worldwide drive to achieve transportation sustainability is public transport, which also plays a vital role in the vibrancy of the economy and supports the general well-being of the population. This concept is an advanced alternative with which to overcome environmental and quality-of-life problems such as air pollution, noise pollution, accidents and traffic congestion [1,2,3,4,5]. The evidence of this is derived from Wall et al. [6], who reported that CO₂ and NO_X emissions per mile, as well as traffic congestion, reduced when people used sustainable modes of transportation, such as public transport in city centres. Further evidence was reported by Replogle and Fulton [7], who predicted a 40% reduction in urban transport emissions by 2050 if people continuously practised the use of public transport, cycling and walking in cities. Reducing the dependency on private transport and enhancing the use of public transport, especially in urban settings, are challenging tasks [8]. Many transportation researchers, policymakers and practitioners have studied the reasons why people take public transport and considered strategies to attract them to choose this as a travel alternative to private transport [2,9,10,11,12].

Nevertheless, the use of public transport compared to private vehicles remains minimal in many parts of the world [13]; for instance, according to Kwan et al. [14], the rate of public transport use in Kuala Lumpur, Malaysia is 17%, compared to the 83% rate for private transport. In addition, Zulkifli et al. [15] mentioned specifically that the rate of rail-based public transport use is one of the most serious concerns in the Asian context. Private transport is preferred over public transport because it is more flexible, comfortable, private and faster, among other benefits [4,16]. Moreover, several studies have reported that the minimal public transport ridership is due to the service quality, which does not appear to meet user expectations [9,17,18,19]. This has led to dissatisfaction with public transport services among their users. Dissatisfied users will neither be loyal to the service nor recommend it to others [20]. Various researchers have investigated how satisfied users are with systems of public transportation, such as railways [8,20,21] and buses [22]. Most indicated that user satisfaction levels are a key feature that motivates their choosing the services again and suggesting their use to other people. High-quality public transport services attract potential users and retain the loyalty of current users. Such high-quality services lead to increased public transport ridership from which service providers can profit.

Travel business organisations must pay particular attention to the satisfaction of their passengers, as focusing on this can be supported by customer-orientated philosophical theory and the key tenets of the continual enhancement that are required by contemporary businesses. Generally, satisfaction is influenced by service quality. Service quality can be defined as the customer’s general evaluation of how the service provider performs [23]. In addition, according to Lai and Chen [8], the extent to which the level of service corresponds to the needs of consumers is measured by service quality. As the work by de Ona et al. [24] indicated, investigations into service quality perceptions do not concur with the essential public transport service quality factors that must be considered. As Table 1 illustrates, these characteristics differ and may include the author, study location or public transport mode.

A growing number of studies have reported the influence of service quality on user satisfaction in several service industries such as healthcare [23,25], aviation [26,27,28], hotel [29,30,31] and cruise [32]. These studies have proved the significant influence of service quality towards users’ satisfaction. For the case of public transport, the degree to which passengers are satisfied is also critically determined by service quality, as evidenced in numerous empirical studies around the world (i.e., [21,22,33]). In addition, recent studies from ASEAN countries such as the Philippines [34,35], Thailand [36,37], Indonesia [38,39] and Malaysia [40,41,42] also proved the significant role of public transport service quality towards passengers’ perceived satisfaction.

In terms of a methodological approach, several models to study the public transport passenger’s perceived satisfaction have been reported in the transportation literature, including the logit model [24,43], the probit model [44,45,46] and the structural equation model [22,33,40,47]. However, these conventional models need assumptions and the pre-requisites. Model assumptions, such as data normality, linear relationship between independent variables and dependent variables, and low multicollinear are rarely observed in consumer satisfaction studies because the findings related to human perception and behaviour (such as passengers’ perceived satisfaction with the monorail, as in this study) are very subjective and tend to have a high degree of heterogeneity [24,43]. In addition, Abbas et al. [48] argued that the conventional paradigm for making predictions, diagnoses and regulations, as well as optimisations such as regression analysis and structural equation models are insufficient when encountering highly complex human and social systems. More recently, the data mining approach such as the artificial neural network (ANN) model has been introduced in behavioural studies to deal with the limitations of the conventional models. The employment of the ANN approach in behavioural related study has shown the improvement in terms of predictive accuracy [49].

The ANN is a non-parametric model that is characterised by its considerable ability to predict and capture highly non-linear intrinsic relationships between variables without the need for the assumptions and pre-requisites often required with other conventional models, such as discrete selection models (e.g., the probit model and the logit model) and structural equation models [49,50]. These arguments support the selection of the methodological approach using the non-parametric model (the ANN, in this study). Thus, taking Kuala Lumpur Monorail as a case study, this study explored the dominant service quality factors of the monorail service that influence passengers’ perceived satisfaction, in order to provide useful information to the service providers, policymakers and planners in formulating effective strategies to increase the monorail service ridership.

The following parts of the current study have been organised into sections as follows: the case study is discussed in Section 2, while the methodology employed is presented in Section 3. Section 4 contains the data analysis results, while the theoretical and practical implications are discussed in Section 5. Finally, the conclusion to the paper is provided in Section 6.

Table 1. Public transport service characteristics.

References		[3]	[5]	[20]	[51]	[52]	[53]	[54]	[55]	[56]	[57]	[58]	[59]	[60]	[61]
Mode of Public Transportation		Bus and Railway	Bus, Tram, Train and Metro	Railway	Bus and Mini Bus	Monorail	Bus	Bus	Bus	Public Transport	Bus	Railway	Railway	Tramway, Metro & Commuter rail	Metro
Service Attributes	Frequency	√	√	√	√		√	√		√	√	√	√	√
	Network coverage					√			√	√			√	√
	Service provision hours	√		√									√	√
	Station parking					√		√				√
	Accessibility		√			√	√				√		√	√	√
	Easy of transfer/Distance									√			√	√
	Ticket price	√	√	√	√	√	√	√	√			√	√	√
	Ticket selling network	√	√	√		√							√	√
	Type of tickets/Passes			√	√								√
	On-board information	√	√	√		√	√	√				√	√	√
	Information at station	√	√	√				√				√	√	√
	Punctuality		√	√	√	√	√	√	√			√	√	√	√
	Access time		√	√			√			√	√		√	√
	Travel speed		√	√			√						√	√	√
	Waiting time								√		√			√	√
	Driver and personnel’s behaviour	√	√	√		√	√	√	√			√	√		√
	Customer service	√			√									√
	Cleanliness		√	√		√	√	√	√		√	√	√	√	√
	Comfort		√	√	√	√	√	√		√	√	√	√	√	√
	Seating capacity		√	√		√				√	√		√	√	√
	Quality of vehicles							√			√	√	√	√
	Temperature			√			√		√					√	√
	Waiting condition				√	√				√	√
	On-board safety	√	√	√	√	√	√	√				√	√	√	√
	Safety at station	√	√	√						√			√	√	√

Note: √ represents that the article contains this type of dimension.

Table 1. Public transport service characteristics.

References		[3]	[5]	[20]	[51]	[52]	[53]	[54]	[55]	[56]	[57]	[58]	[59]	[60]	[61]
Mode of Public Transportation		Bus and Railway	Bus, Tram, Train and Metro	Railway	Bus and Mini Bus	Monorail	Bus	Bus	Bus	Public Transport	Bus	Railway	Railway	Tramway, Metro & Commuter rail	Metro
Service Attributes	Frequency	√	√	√	√		√	√		√	√	√	√	√
	Network coverage					√			√	√			√	√
	Service provision hours	√		√									√	√
	Station parking					√		√				√
	Accessibility		√			√	√				√		√	√	√
	Easy of transfer/Distance									√			√	√
	Ticket price	√	√	√	√	√	√	√	√			√	√	√
	Ticket selling network	√	√	√		√							√	√
	Type of tickets/Passes			√	√								√
	On-board information	√	√	√		√	√	√				√	√	√
	Information at station	√	√	√				√				√	√	√
	Punctuality		√	√	√	√	√	√	√			√	√	√	√
	Access time		√	√			√			√	√		√	√
	Travel speed		√	√			√						√	√	√
	Waiting time								√		√			√	√
	Driver and personnel’s behaviour	√	√	√		√	√	√	√			√	√		√
	Customer service	√			√									√
	Cleanliness		√	√		√	√	√	√		√	√	√	√	√
	Comfort		√	√	√	√	√	√		√	√	√	√	√	√
	Seating capacity		√	√		√				√	√		√	√	√
	Quality of vehicles							√			√	√	√	√
	Temperature			√			√		√					√	√
	Waiting condition				√	√				√	√
	On-board safety	√	√	√	√	√	√	√				√	√	√	√
	Safety at station	√	√	√						√			√	√	√

Note: √ represents that the article contains this type of dimension.

2. The Case Study

The Kuala Lumpur Monorail is an urban rail transit system in Malaysia whose initial purpose was to link many key destinations within central Kuala Lumpur that could not be reached using the other urban rail network systems [52]. A component of the Klang Valley Integrated Transit System, the Kuala Lumpur Monorail was built in 1997 and began operations in 2013. The operator is Rapid Rail, a division of Prasarana Malaysia. It is 8.6 km long and its 11 stations (see Figure 1) are distributed between KL Sentral station and Titiwangsa station. The facility comprises two rails set in parallel for most of its length, apart from a switch to a single-rail arrangement at each end of the line prior to the two final stations. A form of public transport, the monorail uses a single-track (mono) format, with the moving units are located and served by a specific trajectory that hovers in the air [62]. The whole system utilises Malaysian-manufactured trains with two-car trains, each of which potentially accommodates 158 users in normal operating hours. In addition, the Kuala Lumpur monorail system also uses four-car trains to accommodate the passengers, especially during peak hours. At peak times, the service frequency is every seven minutes, while during regular hours it is every ten minutes. The monorail system offers several benefits in comparison to the standard runways that have been constructed, including the need for minimal operational space and reduced interference with the existing flows of traffic, as well as greater cost-effectiveness and time-saving advantages [52,62].

3. Research Methodology

3.1. Questionnaire Design

Prior researchers were the sources from which the author adopted the instrument employed in the current work, including Shen et al. [20], Irtema et al. [3] and Kuo and Tang [63]. Modifications were made to the instrument so that it catered to Malaysia in terms of culture, economy and society, while it was also back-translated into Malay. Before the last stage of gathering information, a pilot test was conducted. Here, 50 respondents from Bandar Baru Bangi, Selangor were chosen at random and issued with copies of the questionnaire. Pilot tests are performed to assess potential weaknesses in the instrument designed for the study, as well as assist in the removal of these issues by the research team before the questionnaire is administered in the full-scale surveying process [64]. The samples from the pilot study provided feedback based on which several questions were removed or modified. Omissions were made since the respondents had not answered them or the responses contained errors. Revisions were made to make the questions clearer and more reliable.

Table 2. Construct items and reliability analysis of the instruments based on pilot study.

Construct	Number of Items	Cronbach’s Alpha (α)
Signage (SG)	5	0.841
Comfort (CF)	5	0.901
Speediness (SN)	4	0.896
Safety (ST)	8	0.813
Ticketing service (TS)	6	0.809
Facilities (FT)	7	0.801
Staff service (SS)	4	0.931
Provision of Information (PI)	4	0.830
Perceived Satisfaction (PS)	4	0.870

shows the pilot study’s reliability analysis results.

The questionnaire used in this study’s final data collection consisted of two main sections: (i) Section 1: Respondents’ demographic characteristics; (ii) Section 2: Service quality and passengers’ perceived satisfaction, as presented in

Table 3. Demographic characteristics.

Variable	Classification	Frequency (n)	Percentage (%)
Gender	Male	212	50.8
	Female	205	49.2
Age (Years)	Less than 20	38	9.1
	21–30	182	43.6
	31–40	156	37.4
	41–50	36	8.6
	More than 50	5	1.2
Education background	Primary school	5	1.2
	Secondary school	30	7.2
	College	68	16.3
	University degree	311	74.6
	Others	3	0.7
Employment status	Full time	233	55.9
	Part time	23	5.5
	Unemployed	24	5.8
	Student	135	32.4
	Others	2	0.5
Monthly income (MYR) MYR1 ≈ USD0.24	Less than 2000	122	29.3
	2001–4000	72	17.3
	4001–6000	121	29
	6001–8000	20	4.8
	More than 8000	4	1
	Private	78	18.7
Driving licence ownership	Yes	356	85.4
	No	61	14.6
Car ownership	0	116	27.8
	1	140	33.6
	2	97	23.3
	More than 3	64	15.3

. A five-point Likert scale was used to measure the Section 2 questions, which ranged from 1 = Strongly disagree to 5 = Strongly agree. Lower scores indicate less agreement with the individual measures.

3.2. Samples and Data Collection

Since the monorail service was operational in Kuala Lumpur, the full-scale investigation occurred in this area. The data in this study were gathered at stations that received the highest numbers of users, including Kuala Lumpur Sentral station and Hang Tuah station. The convenience sampling technique was used to conduct the cross-sectional questionnaire with the respondents. During the data collection process, the target respondents for this study had to satisfy two criteria: (i) to be a Malaysian citizen; (ii) to have experienced using the monorail service within the previous month. Before the survey was administered, each enumerator was provided with a basic briefing on the survey aims, while each potential respondent was requested to state how willing they were to take part. A respondent only received a questionnaire if they expressed a willingness to participate in the survey. The purpose was to ensure that each respondent would accurately and reliably deliver responses to all the questionnaire items. According to Borhan et al. [65], this technique increases the response rate of a survey. A face-to face interview was conducted with each respondent, who also received a small gift after the 10- to 15-min survey to show that their participation was appreciated.

Overall, 500 questionnaires were issued for self-administration between 20 September 2019 and 10 December 2019. The questionnaires that were returned included 83 that were removed since the responses were not valid or/and not complete, leaving 417 questionnaires to be utilised in the subsequent analysis. This resulted in an 83.40% effective rate of response. The characteristics of the 417 respondents are displayed in Table 3.

3.3. Tool and Procedure for Data Analysis

The IBM SPSS Statistics version 24.0 was used to analyse the data in this study, while the respondent demographic profile was analysed using descriptive statistics. The attributes of service quality were extracted into service quality factors using an exploratory factor analysis (EFA). Principal component analysis (PCA) and varimax rotation were the methods used to conduct EFA. Furthermore, the relationship between all the service quality factors and passengers’ perceived satisfaction was evaluated via Spearman’s correlation test. The final stage was to evaluate the dominant factors of monorail service quality that influenced passengers’ perceived satisfaction using the multilayer perceptron neural networks model with a feed-forward backpropagation algorithm. The ANN model was assessed using the sum of squared error (SSE), mean squared error (MSE), root mean square error (RMSE) and the coefficient of determination (R²) value. In addition, the sensitivity analysis was undertaken by determining the normalised importance value.

4. Results

4.1. Exploratory Factor Analysis

Exploration factor analysis (EFA) aims to identify potential latent variable structures, which are then used to reduce variables to smaller and easier-to-manage sizes by removing items that do not have common cores [66,67]. Principal component analysis (PCA) and varimax rotation were the methods used to conduct EFA for the current study, with the aim being to determine the fundamental constructs and dimensions of quality in relation to the Malaysian monorail service. The most frequently recommended technique, this has been widely used in previous studies, especially in the field of transport [63,68,69]. This study maintains the eigenvalues greater than 1 [41,70,71]. The value of eigen is a measure of the degree to which the variance of a variable has been explained by a particular factor. A factor with an eigenvalue assessment ≥ 1 explains more variance than a single variable. Moreover, a factor’s reliability is always positive when it has an eigenvalue that is greater than 1.

EFA was performed on 43 service quality attributes to measure the quality of the Kuala Lumpur Monorail service. The results showed that eight constructs were extracted with eigenvalues greater than 1, and these explained 71.670% of the total variance. As shown in

Table 4. Results of exploration factor analysis.

Factor/Item	EFA
	Loading Factor	Eigenvalue	Explained Variance	Cronbach’s Alpha
Facilities (FT)		23.182	38.450	0.948
Suitable location for vending machines	0.752
Suitable location for the waiting seats	0.795
The compatibility of announcement sound level	0.800
Comfortable waiting seats	0.755
Comfortable armrest and ring set in the train	0.731
Mobile signal strength at the stations	0.704
Mobile signal strength on the train	0.581
Staff service (SS)		2.262	11.689	0.912
Staff appearance	0.677
Staff attitude	0.752
Staff efficiency in resolving passengers’ problems	0.753
Response time of call centre during service hours	0.762
Provision of information (PI)		2.105	5.260	0.930
Publicity of the provided monorail service	0.595
The efficiency of service interruption announcement	0.560
Provision of information on the monorail service at the stations	0.531
Provision of information on the monorail services in the mass	0.543
Ticketing service (TS)		1.706	3.967	0.913
Types of the ticket offered	0.616
Number of ticket vending machines	0.637
Clarity of instruction on using the ticket vending machines	0.610
Self-vending ticket machine works well	0.551
Speed of ticket purchase process	0.589
The convenience of money changing at the station	0.621
Signage (SG)		1.645	3.774	0.915
Signage for the location of the monorail station	0.537
Provision of information at the station	0.527
Signage for automatic gates at the station	0.539
Clarity of the signage for direction	0.512
Speediness (SN)		1.458	3.390	0.921
Punctuality of train arrival	0.666
Acceptable train dwell time	0.614
Acceptable departure interval	0.664
Acceptable service time	0.617
Comfort (CF)		1.175	2.733	0.922
Level of lighting at the station	0.632
Appropriate ventilation and temperature at the station	0.674
Cleanliness at the station	0.709
Appropriate ventilation and temperature in the train	0.739
Cleanliness on the train	0.663
Safety (ST)		1.035	2.407	0.923
Safety at the station	0.518
Safety on the train	0.567
Safety during the travel	0.713
Provision of security alarm facilities	0.710
The behaviour of other passengers	0.773
Advance door closing announcement	0.540
	KMO = 0.969, χ2 = 17,018.015, ρ < 0.000
	Total of variance explained = 71.670

, eight constructs (factors) were labelled as proposed major influences, namely facilities (seven items); staff service (four items); provision of information (four items); ticketing service (six items); signage (four items); speediness (four items); comfort (five items); and safety (six items). In this study, three items were excluded since their loading factors were less than 0.5 and existed in several factors (i.e., they were repetitive) as suggested by Maskey et al. [71] and Uca et al. [72].

After the factors were extracted, the varimax rotation technique which was developed by Kaiser [70] was selected for the EFA protocol, as this method is the most common way to identify key factors and the results are easy to interpret [73,74]. After that, the criteria for conducting EFA were carefully examined. EFA employed three criteria: (i) the Kaiser-Meyer-Olkin (KMO) sampling adequacy and Bartlett spherical test measurements, (ii) each item’s loading factors, (iii) each identified factor’s reliability analysis. As the results reveal, the value of KMO 0.969 demonstrates the high sampling adequacy of the factor analysis process in this study and the Bartlett sphere test found that χ² = 17,018.015, ρ < 0.000 as shown in Table 4 is significant, which explains that the matrix between the correlations has sufficient common variance [74].

In addition, the loading factors for all the measurement items showed that no item had a loading factor lower than 0.5, with all of them ranging between 0.512 and 0.800 (Table 4). Therefore, all the items in listed Table 4 were retained as they complied with the rule of thumb (≥0.5), as proposed by Kuo and Tang [63] and Maskey et al. [71]. Lastly, Cronbach’s alpha values higher than 0.70 form the recommended threshold for the reliability analysis results. The Cronbach’s alpha values for all factors ranged between 0.912 and 0.948 in this study, so these values were in line with the requirements that were proposed by Hair et al. [74]. This shows the reliability of the eight factors that were extracted in this study to assess the quality of the monorail service.

4.2. Spearman’s Correlation Analysis

As shown in

Table 5. Spearman’s correlation coefficients between constructs.

Factor	SG	CF	SN	ST	TS	FT	SS	PI	PS
Signage (SG)	1.000
Comfort (CF)	0.723 **	1.000
Speediness (SN)	0.716 **	0.720 **	1.000
Safety (ST)	0.654 **	0.673 **	0.729 **	1.000
Ticketing service (TS)	0.744 **	0.706 **	0.731 **	0.748 **	1.000
Facilities (FT)	0.641 **	0.670 **	0.635 **	0.564 **	0.681 **	1.000
Staff service (SS)	0.630 **	0.592 **	0.606 **	0.638 **	0.655 **	0.612 **	1.000
Provision of Information (PI)	0.669 **	0.674 **	0.640 **	0.613 **	0.700 **	0.732 **	0.718 **	1.000
Perceived Satisfaction (PS)	0.721 **	0.688 **	0.687 **	0.635 **	0.711 **	0.722 **	0.671 **	0.770 **	1.000

**: p-value < 0.01.

, the results were determined after analysing the Spearman’s correlation between the factors of service quality that were obtained from the exploratory factor analysis (see Table 4) and the perceived satisfaction of passengers. Table 5 indicates that the study variables were positively and significantly related; the study variables featured, for instance, signage, comfort, facilities, speediness, ticketing service, staff service, safety, information provision and perceived satisfaction. Notably, several variables—such as the constructs of (i) comfort and signage, (ii) ticketing service and signage, (iii) speediness and comfort, (iv) ticketing service and comfort, (v) ticketing service and safety, (vi) provision of information and facilities—share a positive and strong relationship (r > 0.7), indicating that the passengers may consider that these issues reflect similar perspectives of perceived quality. On the other hand, all the service quality factors that were considered in this study demonstrated a significantly strong or moderate relationship with perceived satisfaction, indicating that the factors of quality service that were extracted in the previous section represent useful elements of passenger satisfaction.

4.3. Artificial Neural Network Model

In this study, the artificial neural network (ANN) model to predict the influence of Kuala Lumpur Monorail service quality on passengers’ perceived satisfaction was developed using the most frequently used type of ANN model, especially in studies of user behaviour. This is the multi-layered perceptron (MLP) model using the SPSS version 24 [75,76,77,78,79]. The ANN model developed in this study consisted of eight input variables which corresponded to the factors that were obtained through the exploration factor analysis and discussed in Section 4.1. These were facilities (FT), staff service (SS), provision of information (PI), ticketing service (TS), signage (SG), speediness (SN), comfort (CF) and safety (ST). Meanwhile, the output variable was perceived satisfaction.

The possibility of over-fitting was evaded through a ten-fold cross validation, such that the training process utilised 90% of the data, whereas the model testing that was used to measure the prediction accuracy of the training network utilised a further 10% of the data. For an activation function of hidden and output layers, the study utilised the sigmoid function. Moreover, the number of hidden neurons was calculated spontaneously by the neural network module of the SPSS. The structure of the ANN network used here consisted of seven hidden neurons. Thus, the ANN model that was used to predict the service quality factors that influence monorail passengers’ perceived satisfaction had a structural architecture of [8-7-1], as shown in Figure 2.

The ten-fold cross validation was used in this study to prevent the overfitting problem from affecting the model [50,79]. For the current study, it was deemed that in the hidden layer, the seven optimum nods could be regarded as making output variable predictions with accuracy (see Figure 2). The ANN model predictions were estimated for their accuracy using the training and testing data, as applied to the 10 neural networks, with the indicator being the root mean square error (RMSE). For each form of data,

Table 6. SSE, MSE and RMSE value for ANN model.

ANN Network	Training				Testing
	N	SSE	MSE	RMSE	N	SSE	MSE	RMSE
ANN1	372	4.674	0.013	0.112	45	0.650	0.014	0.120
ANN2	375	4.053	0.011	0.104	42	0.442	0.011	0.103
ANN3	374	4.128	0.011	0.105	43	0.456	0.011	0.103
ANN4	363	3.840	0.011	0.103	54	0.689	0.013	0.113
ANN5	365	4.522	0.012	0.111	52	0.402	0.008	0.088
ANN6	377	4.349	0.012	0.107	40	0.298	0.007	0.086
ANN7	372	3.908	0.011	0.102	45	0.656	0.015	0.121
ANN8	376	5.030	0.013	0.116	41	0.144	0.004	0.059
ANN9	369	4.726	0.013	0.113	48	0.667	0.014	0.118
ANN10	377	3.750	0.010	0.100	40	0.842	0.021	0.145
	$\overline{x}$	4.298	0.012	0.107	$\overline{x}$	0.525	0.012	0.106
	SD	0.430	0.001	0.005	SD	0.212	0.005	0.024

Note: ANN: artificial neural network; SSE: sum of squared error; MSE: mean squared error; RMSE: root mean square error.

shows the RMSE in terms of the average values and standard deviations. The results indicated that the average value (training: 0.107, testing: 0.106) and standard deviation (training: 0.005, testing: 0.024) RMSE for the monorail service were relatively low, as shown in Table 6. Therefore, these findings demonstrate that the model for the service has an excellent level of match accuracy.

Furthermore, Leong et al. [50] and Veerasamy et al. [80] mentioned that lower RMSE values indicate that an ANN model has a high predictive accuracy value and that the match between the predicted values and data used would be excellent. The RMSE values of the model developed here were lower than those revealed in reports on the development of the government’s mobile response mobile-system forecast model [78] and the model for determining the smart meter acceptance factor (energy supply) in Malaysia [81]. Moreover, the predictor’s relevance was confirmed by the non-zero synaptic weight quantity associated with hidden neurons.

The current study presents the coefficient of determination (R²) values, as well as the sum of squared error (SSE), mean squared error (MSE) and root mean square error (RMSE) values. The calculated R² values prove that the ANN model that was developed for this study would potentially predict passengers’ perceived satisfaction with the Kuala Lumpur Monorail service with an accuracy level of 79.80%. According to Shen et al. [20], a high R² value indicates a far better match accuracy. Shen et al. [20] also explained that an R² value of 67% is large, one of 33% is moderate and one of 19% is weak. The R² values recorded from the models in this study (≥70%) correspond to those of previous studies that used the non-parametric model ANN when predicting user behaviour [75,76,81].

The following step was sensitivity analysis, defined as each independent factor’s importance in measuring the degree to which the neural network model’s predicted values changed with the independent variable’s values [82,83]. In this study, the level of service factors influencing the passengers’ perceived satisfaction was based on the normalised importance value, as shown in

Table 7. The results of sensitivity analysis.

ANN Network	Relative Importance
	SG	CF	SN	ST	TS	FT	SS	PI
ANN1	0.122	0.053	0.124	0.024	0.089	0.196	0.136	0.256
ANN2	0.181	0.075	0.069	0.002	0.063	0.207	0.134	0.269
ANN3	0.074	0.091	0.130	0.055	0.090	0.174	0.075	0.311
ANN4	0.172	0.035	0.091	0.006	0.077	0.213	0.109	0.297
ANN5	0.160	0.133	0.115	0.035	0.136	0.156	0.043	0.223
ANN6	0.127	0.061	0.131	0.005	0.090	0.209	0.117	0.259
ANN7	0.164	0.071	0.057	0.012	0.086	0.193	0.149	0.267
ANN8	0.138	0.140	0.137	0.093	0.096	0.185	0.086	0.124
ANN9	0.116	0.075	0.087	0.058	0.093	0.184	0.111	0.275
ANN10	0.167	0.065	0.076	0.014	0.149	0.164	0.123	0.243
Average of relative importance	0.142	0.080	0.102	0.031	0.097	0.188	0.108	0.252
Normalised relative importance (%)	56.3	31.6	40.2	12.1	38.4	74.5	42.9	100.0
Rank	3	7	5	8	6	2	4	1

Note: ANN: Artificial Neural Network; SG: signage; CF: comfort; FT: facilities; SN: speediness; TS: ticketing service; SS: staff service; ST: safety; PI: provision of information.

. This value can be derived by dividing the importance value for each independent variable (input neuron) by the greatest importance value; it is shown as a percentage [79,84]. The sensitivity analysis results showed that the three most influential factors in predicting passengers’ perceived satisfaction with the monorail service (Table 7) were the provision of information (100.0%), facilities (74.5%) and signage (56.3%). In addition, the three factors with the lowest influence on passengers’ perceived satisfaction with this service were ticketing service (38.4%), comfort (31.6%) and safety (12.1%).

5. Discussion

The findings of the current research offer in-depth and extensive details about passenger perceptions of the Kuala Lumpur Monorail service quality. This study indicates the ranking of the service factors based on their respective contribution to passengers’ perceived satisfaction with the quality of the monorail service provided. The approach used in this ranking process was based on the non-parametric modelling technique, the artificial neural network (ANN) model.

Eight factors would potentially affect the perceived satisfaction of passengers, and these were used with measurement scales in the factor analysis. The constructs were signage, comfort, facilities, speediness, ticketing service, staff service, safety and information provision. Identical constructs were employed in a Chinese context by Shen et al. [20], who examined how satisfied users of the Suzhou urban rail transit service were. Additionally, factors resembling these were utilised by de Oña et al. [85] and Yanık et al. [61], each of whom undertook a case study (in Italy and Turkey, respectively) and included comfort, security, information provision, customer service and other factors. Based on previous studies, the factors extracted from the exploratory factors analysis were those that reflected service quality and seemed to represent dimensions underlining the passengers’ perceptions of quality. These factors have the potential to influence the perceived satisfaction of Kuala Lumpur Monorail users. The indicator factors explaining the majority of variance in the current work were facilities, staff service and provision of information.

Additionally, the researchers investigated the Kuala Lumpur Monorail service to ascertain how the perceived satisfaction of passengers was affected by service quality factors. In this study, the effects of a factor (presented as a ranking) were determined based on the relative contribution (the value of normalised relative importance) these factors made to passengers’ perceived satisfaction. The approach used in this identification of the ranking process was the ANN model. This technique was used due to its various advantages (as reported in previous studies) over conventional parametric model methods such as the regression model, structural equation model or logit/probit model [49,77]. In addition, the ANN technique has also been reported to be more effective than other non-parametric models such as the decision trees technique [49]. Proof of this is based on the predictions reported for the model in this study (79.70%) being more accurate than those of the result tree method (between 59.72% and 62.16%) in assessing the relative contribution of the public transport service quality, as reported by de Oña et al. [86].

Based on the percentage of the relative importance value and the monorail service quality ranking identified here, three factors—provision of information, facilities and signage—were found to be important in this assessment of the passengers’ perceived satisfaction with the services provided. On the other hand, the comfort and ticketing service factors were considered to be less influential on the passengers’ perceived satisfaction with the KL monorail service. The validity of these methodologies and findings was confirmed through the parallel results that were reported by previous studies in the transport field, which have used different techniques such as importance-performance analysis, structural similarity models and regression models [21,52,61,87].

The results of the current work suggest that the perceived satisfaction of passengers could be increased through an alternative: monorail stations and train carriages could display correct, dependable and updated information. As van Lierop et al. [13] and Machado-Leóna et al. [60] recommended, various types of information should be provided by the service operators and responsible authorities because such details are crucial to improving the levels of perceived passenger satisfaction with public transport. This information should cover ticket fares, interruptions to services (due to disruption), service timings, train routes and arrival/departure timetables. This is because such information is highly important for passengers when planning and managing their journeys.

In terms of facilities, monorail service providers must provide adequate quality and facilities on trains and at stations. For instance, while they wait for trains or travel on the services, passenger comfort could be improved by providing sufficient comfortable seats at the station and on the train. According to Gao et al. [88], comfortable seats can increase passenger satisfaction with public transport. In addition, the service provider must ensure that holders for standing passengers (such as hanging straps, grab handles, handrails and stanchions) are sufficiently convenient. Moreover, installing clear and systematic signage is important for increasing passengers’ perceived satisfaction, especially among new riders.

Other factors that cannot be overlooked are comfort and customer service, which were found to have significant impacts on passengers’ perceived satisfaction in previous research. In this study, the comfort factor was influenced by temperature and cleanliness. According to Geetika [89] and Ibrahim et al. [18,90], service providers should ensure the comfort of the passengers while on the train and at the station as this can significantly influence the passengers’ perceived satisfaction. The temperature and cleanliness of trains and stations also contribute to the comfort factor. Given that the monorail service studied here is a form of urban rail transit, the cleanliness of its facilities could be enhanced by providing additional and better-positioned bins to make it more convenient for passengers to throw away their litter. In addition, the provision of recycling bins is also an important aspect of efforts to improve the cleanliness of the facilities, while it would also encourage recycling activities and contribute to a greener environment (sustainability) [18]. In addition, policies that prohibit smoking, eating or drinking on trains also contribute to improving the cleanliness and comfort of passengers when using the monorail facilities [91]. In this regard, passengers can also contribute to ensuring that monorail services meet their expectations by complying with the cleanliness rules and policies.

Staff service was also one of the key factors identified in this study that would potentially influence passengers’ perceived satisfaction with monorail services in Kuala Lumpur. Management teams, drivers and other staff must contribute collectively to ensure the passengers are satisfied with their services. The monorail management must ensure that their employees represent the company and business positively. While engaging with a passenger, staff members need to display courtesy and professionalism. They also need to supply correct, updated and dependable details, which is particularly the case with customer service counter employees. Furthermore, the appearance of the workforce would be improved with professional uniforms [18].

Overall, based on the findings obtained in this study, the measures outlined are important for increasing passengers’ perceived satisfaction with monorail services in Malaysia. This knowledge will contribute to efforts to increase the usage of urban rail transit services and reduce the public’s dependency on private vehicles for urban travel, especially in Kuala Lumpur.

6. Conclusions

The current study reported the dominant service quality factors influencing passengers’ perceived satisfaction with the monorail service in Kuala Lumpur, Malaysia. To summarise, the research findings reveal that the perceived satisfaction of passengers could be affected by eight factors: signage, comfort, speediness, safety, ticketing service, facilities, staff service and information provision. In addition, these monorail service quality factors demonstrated a positive and strong relationship with passengers’ perceived satisfaction. Furthermore, the findings obtained from the non-parametric model of the artificial neural network indicated that three factors—provision of information, facilities and signage—were dominant in terms of influencing the riders’ perceived satisfaction with this Malaysian monorail service. This can be explained by the significant contribution these factors made to forming the passengers’ perceived satisfaction with the service provided. The contribution of these findings should facilitate improvements to aspects of both the theory and the practice. The current work should also assist service providers, policymakers and academics to identify and specify which useful approaches should be introduced to enhance monorail services. Consequently, passenger satisfaction will improve in the short term and thereafter increase the ridership. Subsequently, the service providers’ profits will be maximised, enabling the transportation market to be sustainable in the long term.