Assessment of drivers and barriers in the adoption of Mobility as a Service (MaaS): a case study of Noida, India

ABSTRACT

Mobility as a Service (MaaS) has emerged as a transformative concept in urban transportation, integrating multiple modes of transportation through digital platforms to provide seamless travel experiences. By utilizing smartphone applications, MaaS simplifies trip planning, real-time information access, and consolidated payment systems, offering convenience to users. However, MaaS research has primarily focused on developed countries with well-established transport systems, making it crucial to explore its potential and challenges in developing cities of the global south, such as Noida in India. In this study, Noida, a satellite town of NCT Delhi, was chosen as only case study to gather data on user behaviour and preferences regarding app-based mobility services. A comprehensive survey collected information on socio-economic factors, personal vehicle ownership, commuting patterns, public transport usage, and attitudes towards digital ecosystems and app-based mobility services. Principal Component Analysis and K-means Clustering techniques were applied to identify distinct user types and categories, providing insights into user preferences and expectations. The analysis of the collected data revealed user clusters and their respective characteristics. The sustainability aspects of on-demand mobility services were evaluated, comparing user perceptions with private vehicle usage. The study also examined the impact of app-based mobility services on public transport and identified barriers and constraints specific to different user clusters, contributing to a better understanding of the feasibility of implementing MaaS. The findings will provide valuable insights for policymakers and transportation authorities, enabling the development of strategies and interventions to enhance urban mobility and foster MaaS adoption. By addressing the specific needs and preferences of users, MaaS can play a significant role in improving the efficiency and sustainability of transportation in complex urban environment cities like Noida in India.

KEYWORDS:

• Mobility as a Service (MaaS)
• user behavior
• app-based mobility services
• sustainability
• principal component analysis (PCA)
• K-means clustering

1 Introduction

1.1 Background

Mobility as a Service (MaaS) is a user-focused approach to transportation that enables users to conveniently select and access different modes of transport through a digital platform. By integrating public transport and shared mobility services, MaaS aims to create a connected, convenient, and sustainable urban transport ecosystem while reducing reliance on private vehicles. In India, the mobility landscape is characterized by a multitude of public and private actors operating in an unorganized manner, impacting reliability, efficiency, and user experience. The successful implementation of MaaS requires addressing supply-side challenges, such as operator acceptance of digital systems, and understanding demand-side factors, including user behavior and willingness to shift from private vehicles.

Information and Communication Technologies (ICT) have significantly transformed travel patterns and preferences, offering real-time traveler information, online shopping reducing physical travel, and the emergence of app-based mobility services. MaaS leverages these advancements by integrating multiple transport modes into a single platform, providing seamless and convenient travel options. Successful MaaS implementations in developed countries, such as Whim in Helsinki and Ubigo in Gothenburg, showcase the potential of integrated mobility solutions. In India, ride-hailing services like Ola and Uber have gained popularity as alternatives to unorganized public transport and private vehicle ownership, indicating the market’s readiness for MaaS.

Public transport integration is a vital component of MaaS, offering benefits such as increased ridership, cost savings through integrated fare systems, and environmental sustainability by reducing emissions and congestion. Public transport within a MaaS framework enhances accessibility, sustainability, and efficiency in transportation systems. However, concerns about the sustainability of app-based cab services and their impact on public transport modes and greenhouse gas emissions have been raised. The Indian government has implemented various policies and initiatives to promote sustainable transportation, including the National Urban Transport Policy and the Electric Mobility Mission Plan. To foster the adoption of MaaS, the Ministry of Housing and Urban Affairs is working on establishing smartphone-enabled connected mobility platforms to integrate shared transportation services and provide route options and payment gateways.

Exploring Mobility as a Service (MaaS) in Indian cities is crucial due to the unique challenges and opportunities presented by the urban transport landscape (Mitra et al., 2021). MaaS has the potential to improve accessibility, reduce congestion, and promote sustainable travel options. However, notable gaps exist in the literature. Limited research has been conducted on user attitudes towards smartphone applications for trip planning and travel-related purposes, which are integral to MaaS implementation. Further investigation is needed to understand user preferences, concerns, and behavioral patterns to design effective MaaS systems for Indian users (Mitra et al., 2021). Additionally, there is a lack of studies focusing on the sustainability aspects of app-based mobility services, necessitating research on environmental impacts. Furthermore, research on the integration of Public Transport within the MaaS framework in Indian cities is limited, requiring investigation into user perceptions and willingness to adopt shared mobility options). Addressing these gaps through further research will provide valuable insights for policymakers and transportation planners, enabling the development of user-centric, sustainable, and efficient MaaS systems tailored to Indian urban environments.

1.2 Aim and objectives

Aim. This study aims to address some of these research gaps in the context of Mobility as a Service (MaaS) in India. Specifically, it focuses on understanding user attitudes towards smartphone applications for trip planning and travel-related purposes, their perceptions of the Public Transport system, and their usage and attitudes towards on-demand app-based mobility services.

Objectives

1. Develop the user clusters based on their attitudes and preferences towards smartphone-enabled digital ecosystems, public transport services, and on-demand mobility services as these three are vital for developing an effective MaaS system.

2. Identify the variation in mobility pattern and choice across identified user clusters.

3. Cluster wise variation in response on impact of digital ecosystem and on demand mobility services on private vehicle use and ownership, essentially focusing on sustainability aspect of a full scale MaaS deployment.

The findings will offer valuable insights to policymakers and transportation authorities, facilitating the development of strategies and interventions to enhance urban mobility and promote MaaS adoption. By addressing users’ specific needs and preferences, MaaS can play a significant role in enhancing transportation efficiency and sustainability in Indian cities.

The remaining sections of this paper are structured as follows. Section 2 provides an in-depth review of the relevant literature that forms the basis for this study. In Section 3, a concise overview of the case study is presented. Section 4 outlines the methodology employed in the study. The survey design and data collection process are detailed in Section 5. Section 6 explains the data analysis procedures undertaken. In Section 7, the study’s methods and results are discussed. Finally, the paper concludes with a summary of findings and conclusions in the final section.

2 Literature review

MaaS (Mobility as a Service) is a transformative concept in transportation that aims to integrate various modes of transport into a user-centric platform (Jittrapirom et al., 2017). It seeks to provide individuals with a comprehensive and personalized mobility experience by combining public transit, shared mobility services, cycling, walking, and on-demand solutions. User-friendly mobile applications or digital platforms play a crucial role in realizing MaaS by enabling users to plan their journeys, access real-time information, and make seamless ticketing and payment transactions. Successful implementation of MaaS requires collaboration among public transport operators, private mobility service providers, technology companies, and local authorities to create a connected and interoperable transport network.

Cities worldwide have embraced MaaS through pilot projects, with notable examples including Helsinki, Singapore, Vienna, and London. Helsinki’s Whim initiative has been at the forefront, offering a pioneering package of mobility services through a single application. This initiative integrates public transport, taxis, car rentals, and bike-sharing, providing users with a comprehensive range of options for their travel needs. These pilot projects demonstrate the potential of MaaS to revolutionize urban transportation by providing users with convenient, sustainable, and integrated mobility solutions.

The objectives of MaaS are multifaceted. Firstly, MaaS aims to enhance the travel experience by simplifying journey planning, reducing the need for multiple apps or ticket purchases, and offering personalized recommendations based on user preferences and real-time conditions. Secondly, MaaS promotes sustainable mobility by encouraging the use of shared and environmentally friendly modes of transport, thereby reducing reliance on private car ownership. By addressing traffic congestion, improving air quality, and reducing carbon emissions, MaaS contributes to creating more sustainable and livable cities 2020. Additionally, MaaS strives to enhance the integration and coordination of transport services, enabling seamless transitions between different modes of transport with a single ticket or payment system. This optimization of existing infrastructure and resources improves overall transportation efficiency and promotes a more equitable and accessible transport network (Jittrapirom et al., 2017).

Smartphone applications play a vital role in enabling MaaS by providing real-time information on various transportation options, facilitating informed decisions about routes and modes of transportation 2022. Additionally, these apps have contributed to the rise of on-demand transportation services like ride-hailing platforms, reducing the need for personal car ownership and usage, resulting in decreased traffic congestion and parking demand 2020. Smartphone apps have also promoted active modes of transportation like walking and cycling by integrating bike-sharing systems 2022. They have further facilitated multimodal journeys, allowing users to plan and pay for trips involving a combination of different modes within a single app. As a result, smartphone applications have transformed urban travel patterns, promoting sustainability and efficiency in urban transportation systems.

Several studies have examined the relationship between smartphone app usage, socio-demographic variables, and travel outcomes of users. Gupta and Sinha (2022) employed Latent Class Cluster Analysis to classify users into clusters based on their app usage, private vehicle usage, intermediate public transport usage, and public transport usage. Their study found that younger users with higher education, more smartphone experience, medium-to-high household income, and lower vehicle ownership had a high probability of being classified as multimodal travelers. Another study by Jamal and Habib (2019) conducted in Canada revealed that smartphone use frequency for communicating and coordinating trips with others was positively associated with pro-environmental statements related to GHG emission fines. The study also found a significant influence of age on smartphone use for travel, with millennials being more likely to use smartphones for trip planning activities and experiencing increased travel outcomes.

Despite the potential benefits of MaaS, there are barriers to its widespread adoption 2021. These include transport operators’ unwillingness to use third-party payment systems or allow third parties to sell their tickets, the requirement for a single identity for travelers across modes, and concerns about privacy and security of personal information. Users may also be hesitant to adopt MaaS if they perceive it as more expensive than their current mode of transportation or if they prefer the flexibility of owning their own vehicle. Accessibility of the platform may pose challenges for some users, particularly those with limited digital literacy or older individuals 2022.

Studies have also explored user behavior and preferences related to ride-hailing services in developing countries 2019. Factors influencing the demand for ride-hailing services were assessed, including socio-demographic variables, trip features, cost, and service attributes like comfort and reliability (Raj et al., 2023). The impact of ride-sharing on public transport usage has been examined in studies conducted in the US, the Netherlands, and Canada, highlighting how ride-hailing services can both compete and complement public transport (Cats et al., 2021; Tarnovetckaia & Mostofi, 2022; Zhang & Zhang, 2018). These studies emphasize the need for integrating different modes of transport to enhance sustainable mobility.

Overall, MaaS offers the potential to provide users with a convenient and efficient transportation experience by integrating various modes of transport into a user-centric platform. While smartphone applications play a crucial role in enabling MaaS, there are barriers to widespread adoption, including resistance from transport operators, concerns about privacy and security, and user preferences for flexibility and cost-effectiveness. Understanding user behavior and preferences, as well as addressing these barriers, are essential for the successful implementation and promotion of MaaS as a sustainable and efficient transportation solution.

3 Case study profile – Noida

Noida, located adjacent to Delhi in the National Capital Region (NCR) of India, has been chosen as the study area. It covers an approximate area of 203 square kilometers and had a population of over 0.637 million people according to the 2011 census. Noida holds a prominent position in terms of economic output among major cities in India. The town is renowned for its robust industrial and commercial sectors, particularly in information technology, manufacturing, and services. The city’s economic significance is evident from its high ranking in terms of GDP contribution, consistently being one of the top contributors to the economy of the state of Uttar Pradesh. The strategic location of Noida, coupled with well-planned infrastructure and flourishing industries, has been instrumental in driving its economic growth and attracting both domestic and international companies. The presence of business parks, industrial zones, and commercial centers in Noida has generated employment opportunities and fostered overall economic development.

4 Methodology

The study comprises three essential components: a background study and research question formulation, data analysis and modeling, and result interpretation and discussion. The initial phase involves conducting a thorough background study, defining the research question, and collecting relevant data. The next stage employs sophisticated techniques to analyze and interpret the collected data, extracting meaningful insights and drawing accurate conclusions. Finally, the study concludes with result interpretation and discussion, where the findings are scrutinized, compared with existing literature, and interpreted within the context of the research question. The methodology employed for this study is outlined in the illustrative framework as Figure 1, providing a step-by-step outline of the entire research process.

Figure 1. Research methodology.

5 Survey design & data collection

The study utilized a Google form consisting of 51 questions divided into 27 subheads, which was distributed through social media applications and employed the snowballing technique. A total of 331 valid samples were collected between February and April 2023. The questionnaire was structured into five sections as follows:

1. User’s socio-demographic background: This section aimed to gather general profiling information about the respondents, including their age, gender, income, education, and occupation. The collected data were recorded using categorical choices.

2. Vehicle ownership and usage pattern: Participants were asked about the number of personal vehicles they owned, including cars and two-wheelers. The frequency of usage for different purposes was also recorded using a likert scale.

3. Daily commuting characteristics & Public Transport usage: This section captured respondents’ daily commute to work or education, focusing on the first mile, main haul, and last mile segments. Trip cost, time, and length were grouped into ranges using an ordered scale. Additionally, users’ preferences for selecting their mode of commute were measured by assessing their agreement with specific statements in a likert scale format. The purpose-wise frequency of public transport usage and the respondents’ attitudes towards specific aspects of public transport, influencing their decision to not use it, were also recorded.

4. Digital Technology usage: This section examined the frequency of smartphone application usage for digital payments and trip planning purposes. It also gathered attitudinal data regarding users’ concerns within a smartphone-based digital ecosystem.

5. Preferences and usage pattern of Ride-Hailing Services: This section collected data on the frequency and purpose of using ride-hailing services, along with the factors driving their adoption. Respondents were also asked to identify their next best alternative mode in place of ride-hailing services, providing insight into the modes that ride-hailing services have replaced. The stated impacts of ride-hailing services and the digital ecosystem on personal vehicle usage and the purchase decision of new vehicles were recorded.

For all ordered responses, a 4-point likert scale was adopted, although a 5 or 7-point scale is generally recommended. This decision was made to enhance user comfort and the likelihood of answering the entire set of questions, as initial users reported longer time requirements for survey completion. A similar likert scale has been utilized in a research study conducted in Germany for concerning mobility behavior (Krauss et al., 2023). Frequency data responses were recorded as ‘Never’, ‘Sometimes’, ‘Often’, and ‘Always’, while attitudinal data responses indicated the users’ level of agreement with specific statements as ‘Strongly disagree’, ‘Somewhat disagree’, ‘Somewhat agree’, and ‘Strongly agree’.

The combined surveys provide a thorough insight of the intentions and actions of users in relation to Mobility as a Service (MaaS). Using a structured questionnaire, this study explores various adoption and usage patterns of MaaS. The first component focuses on socio-demographics, which are vital for evaluating the possible user base and tailoring MaaS offers. The second segment explores vehicle ownership, offering light on the potential for MaaS to replace or supplement individual automobiles. The third portion examines daily commuting and public transportation usage, demonstrating how MaaS could be integrated into existing routines. Important considering MaaS’s emphasis on digital platforms, the fourth segment addresses digital technology usage. The fifth segment concludes by discussing the influence of MaaS on ride-hailing services and other transportation modes. This aggregate data supports the development, deployment, and modification of MaaS, guiding strategies for promoting adoption and adapting services to the different demands of users.

6 Data analysis

6.1 Preliminary statistical tests

This research entails the collecting and analysis of a vast number of variables with varied degrees of relevance. For further investigation, it is crucial to identify the most important and significant of these variables.

The Likert scale data and ordered data have been subjected to the Cronbach’s alpha reliability test. Cronbach’s alpha quantifies the degree to which Likert scale items are interrelated and give a reliable estimate of a single underlying dimension. A high Cronbach’s alpha value suggests that the questionnaire items are highly associated and measure the target construct in a consistent manner. A low Cronbach’s alpha score may indicate, however, that the questionnaire items are measuring various constructions or are not reliable measures of the targeted construct. Alpha values greater than 0.7 are often advised. The following Table 1 displays Cronbach’s alpha for frequency-based and attitudinal Likert scale data. It is determined that responses are sufficiently dependable.

Table 1. Cronbach’s alpha value.

Frequency data	Alpha	Attitudinal data	Alpha
Private vehicle usage	0.74	Commute preferences	0.73
Public Transport usage	0.81	PT attitude	0.7
Smartphone App usage	0.76	Digital ecosystem concerns	0.69

The chi-squared test was used to analyze the relationship between socio-demographic variables and the frequency of mode usage and smartphone application usage for trip planning purposes. This statistical method determines if there is a significant association between two sets of variables. It compares observed frequencies in a contingency table to expected frequencies based on the assumption of independence. The test calculates the chi-squared statistic, which measures the difference between observed and expected frequencies, considering the sample size. By comparing this statistic to the critical value from the chi-squared distribution, one can determine if the association between variables is statistically significant or if any differences observed are likely due to chance. This test is widely used in various fields, such as social sciences, healthcare, and market research, to gain insights into the relationships between categorical variables. Table 2 presents the significance levels of each socio-demographic variable for mode usage and app usage frequency. Further, Figure 2 also shows the distribution mode usage frequency of respondents based on their income and education.

Table 2. Test of association – significance levels.

Mode use frequency	Income	Age	Gender	Education	Occupation
Car - work purpose	0.004	0.050	0.037	0.115	0.156
Car - shopping	0.001	0.007	0.004	0.192	0.457
Car - recreational	0.054	0.001	0.034	0.079	0.266
Car - airport/station	0.002	0.012	0.068	0.142	0.147
Car - other	0.097	0.105	0.062	0.218	0.111
2W - work	0.001	0.006	0.545	0.272	0.035
2W - shopping	0.005	0.174	0.698	0.166	0.101
2W - recreational	0.011	0.366	0.739	0.057	0.276
2W- airport/station	0.001	0.457	0.275	0.162	0.111
2W - other	0.057	0.562	0.694	0.374	0.05
PT - work	0.145	0.131	0.016	0.105	0.174
PT - shopping	0.293	0.002	0.038	0.207	0.017
PT - recreational	0.079	0.009	0.071	0.011	0.105
PT - airport/station	0.175	0.014	0.065	0.091	0.142
PT - other	0.297	0.017	0.001	0.257	0.005
Ride hailing usage	0.051	0.251	0.288	0.013	0.001
Smart Phone App usage	0.011	0.002	0.154	0.051	0.145

Figure 2. Mean score for purpose wise mode usage frequency for age & income.

The analysis revealed that Income and Age exhibited a stronger association with mode usage frequency. In other words, these socio-demographic variables had a more significant relationship with the frequency of using different transportation modes. On the other hand, when considering smartphone application usage for trip planning purposes, Age emerged as the variable with a stronger association compared to other socio-demographic factors. This implies that Age played a more prominent role in determining the frequency of using smartphone applications for trip planning purposes, compared to other variables in the study.

6.2 Principal component analysis

The study employed Principal Component Analysis (PCA) to identify distinct latent variable constructs related to behavioral components such as commute preferences, perception of ride-hailing services, and digital ecosystems. With a large number of latent variables (24 in total), a data reduction technique was utilized to make the analysis more manageable. PCA is a statistical method that identifies patterns and reduces the number of variables in a dataset. It transforms a large set of variables into a smaller set of uncorrelated variables called principal components. The goal of PCA is to extract the most important information from the original variables and reveal the underlying structure of the data. PCA is widely used in various fields to reduce dimensionality and identify crucial features.

The data was appropriately prepared so that all responses measured a particular construct consistently. The questionnaire was designed with a focus on respondent understanding rather than uniform measurement of constructs. For statistical analysis, the collected data was reordered to measure constructs consistently, which is a necessary condition for reliability tests and multivariate analysis like PCA.

The selection of input variables for PCA included attitudinal and preference variables measured for each questionnaire component. An important criterion for PCA is having a good correlation between input variables without excessive closeness to 1. To satisfy this condition, several permutations and combinations were explored, starting with 24 input variables and progressively reducing to 21. The input variables underwent the Kaiser-Meyer-Olkin (KMO) and Bartlett’s test of sampling adequacy, which indicated a good sample size (KMO score of 0.8), significant correlation matrix (significance level < 0.05), and acceptable collinearity (determinant of 0.002).

Component rotation is a crucial step in PCA to enhance the interpretability of results. While PCA generates uncorrelated components based on variable variances, these components may lack straightforward interpretation. Rotation techniques, such as Varimax, Quartimax, or Oblimin, aim to simplify the component structure by maximizing the variance of some components while minimizing the variance of others. Rotated components represent more meaningful constructs or factors that are easier to understand and interpret. Rotation clarifies relationships between variables, revealing the underlying structure of the data. In the current analysis, the Oblimin rotation technique resulted in a better interpretation of components. The extracted components, their loadings, and their explanatory power (explaining 60.7% of variance) are presented in the Table 3. The questionnaire demonstrated high reliability (Cronbach’s alpha > 0.7) and good internal consistency of the scale (standard factor loadings > 0.5). Convergence validity and discriminant validity were evaluated to assess the questionnaire’s validity.

Table 3. Rotated component Matrix.

	Components
Items/Measures	1	2	3	4
Ease of travel and comfort drives my use of ride hailing services	0.801
I prefer using ride hailing over Public Transport regardless of connectivity	0.678
prefer using ride hailing if they are available at reasonable fare	0.763
High parking charges drives me to use ride hailing services	0.723
I do not prefer travelling in public transport because of crowding	0.863
Convenience, avoiding waiting and negotiation is what drives my usage of ride hailing services	0.912
Prestige/status in using Public Transport is major issue for me			0.851
Public Transport is often unclean and lacks hygiene			0.777
Public transport is always overcrowded			0.763
Public Transport has last mile connectivity issues			0.546
Public transport is unsafe to travel			0.621
Public transport lacks privacy			0.685
I am ready to spend more on commute if it satisfies my requirements				0.789
I would prefer a mode which saves time in daily commute				0.864
Comfort and ease of travel is very important in daily commute				0.908
Safety and security is an important criterion for daily commute				0.862
I value privacy when choosing mode of commute				0.774
Reliability of apps is an issue in a digital ecosystem		0.768
There is possibility of fraud in a digital ecosystem		0.755
I am worried about the over reliance of digital ecosystem		0.934
I am worried about the data privacy of digital ecosystem		0.912

The extracted components are highly interpretable. Component 1 can be understood as representing the factors that drive users’ behavior in using ride-hailing services. Component 2 captures the level of digital consciousness among users. Component 3 explains the factors behind users’ low usage of public transport. Lastly, component 4 May represent the factors that influence users’ mode preferences. The factor scores or component scores derived from these four components for each sample serve as inputs for the subsequent clustering analysis.

6.3 K mean clustering

The utilization of K-means clustering on component scores derived from Principal Component Analysis (PCA) offers an effective approach for exploratory data analysis and pattern recognition. By using component scores instead of the original variables, this methodology combines the advantages of dimensionality reduction and clustering to unveil underlying structures in high-dimensional datasets. The process begins with PCA, which identifies the principal components capturing the most significant variation in the data. The component scores, representing the projections of the original variables onto these components, are then extracted. These scores provide a lower-dimensional representation where each observation is represented by its coordinates along the principal components. K-means clustering is subsequently applied to the component scores, partitioning the data into K distinct clusters based on similarity.

By clustering on the component scores rather than the original variables, this approach reduces dimensionality, potentially enhancing the clustering performance and interpretability of the results. It effectively addresses the curse of dimensionality by focusing on the most informative components and removes the influence of irrelevant or noisy variables, leading to more robust clustering outcomes. Moreover, using component scores promotes better interpretability and comprehension of the clustering results. This combined PCA and K-means clustering methodology offers several advantages, allowing researchers to identify homogeneous groups or patterns within complex datasets. It serves as a powerful tool for unsupervised analysis, enabling the extraction of valuable insights from the data.

After specifying the convergence criteria in the k-means clustering algorithm, the number of clusters was iteratively reduced, starting from an initial higher number of 6 clusters. To determine the stable number of clusters and validate the extracted clusters, the means of the clusters were compared using an analysis of variance (ANOVA). ANOVA allows for a statistical evaluation of whether the means of different clusters exhibit significant differences, providing insights into the quality and reliability of the clustering solution. The validation process typically involves applying the k-means algorithm to assign observations to clusters, calculating the cluster means, and conducting ANOVA to test the null hypothesis of no significant mean differences among the clusters. By examining the variation between and within clusters, ANOVA determines the statistical significance of the observed mean differences. The utilization of ANOVA to validate k-means clustering enhances the robustness of the results, providing a statistical basis to assess the distinctiveness and homogeneity of the clusters and gaining insights into meaningful groupings. In this study, starting with 6 clusters and progressively reducing to 3 clusters yielded a stable solution using the validation method. Further the distribution of samples in the cluster are shown in Figure 3.

Figure 3. Extracted clusters.

The extracted clusters were plotted onto the 4 components extracted from the principal component analysis to assess how they vary along the 4 components. Parallel coordinates plots are highly useful for representing multidimensional data. They provide a compact and intuitive visualization by using parallel axes to display multiple variables simultaneously as shown Figure 4. Parallel plots enable the detection of relationships, identification of outliers, and exploration of data across dimensions. In this study, the three extracted clusters were depicted on the four principal components through this intuitive visualization technique, enabling the detection of relationships.

Figure 4. Extracted clusters represented over 4 principal components.

6.4 Cluster analysis results

The three clusters identified have reasonable distribution without excessive skewness, the largest cluster is cluster 1 accounting for 38% of the cluster, while cluster 2 is the smallest cluster accounting for 28% of the sample set. The socio demographic characteristics of these clusters has been briefly discussed as shown in Table 4.

Table 4. Socio demographic distribution of clusters.

Parameter	Component	cluster 1	cluster 2	cluster 3
Gender	Male	44%	33%	23%
	Female	30%	22%	48%
Age	<18 years	0%	50%	50%
	18–29 years	25%	32%	43%
	30–44 years	47%	30%	23%
	45 to 60 years	53%	14%	33%
	>60 years	50%	0%	50%
Education	Primary Level	14%	81%	5%
	Intermediate Level	18%	50%	33%
	Diploma	29%	39%	32%
	Graduation	38%	28%	35%
	Post graduation & Above	51%	6%	43%
Occupation	Business/self employed	60%	33%	7%
	Government Service	41%	23%	36%
	Private service	42%	28%	31%
	Student	15%	30%	55%
	Homemaker	0%	38%	63%
Income	<10,000	0%	30%	70%
	10,000–25,000	5%	43%	51%
	25,000–50,000	5%	39%	56%
	50,000–1,00,000	72%	20%	8%
	>1,00,000	98%	0%	3%
car ownership	0	10%	34%	56%
	1	43%	30%	26%
	2	94%	6%	0%
	3	90%	10%	0%
2W ownership	0	35%	24%	41%
	1	31%	35%	35%
	2	57%	20%	23%
	3	57%	29%	14%

Cluster 1 consists of individuals with a higher proportion of males (44%) compared to females (30%). The majority of individuals in this cluster fall into the age range of 45 to 60 years (53%). Educationally, there is a diverse distribution, with a significant portion of individuals having attained a doctorate degree (100%). Occupationally, this cluster is predominantly composed of business/self-employed individuals (60%). In terms of income, a significant proportion of individuals in this cluster have higher income levels, with a majority earning above 1,00,000 (98%). They also exhibit the highest car ownership percentage (56%) and relatively high two-wheeler ownership (41%). This cluster can be termed as affluent professional class.

Cluster 2 displays a relatively balanced gender distribution, with males accounting for 33% and females for 22%. The age distribution in this cluster is relatively even, with the highest proportion falling in the 30 to 44 years range (32%). Educationally, there is a diverse distribution, with the highest proportion having a diploma (39%). Occupationally, this cluster has a higher proportion of individuals in government service (36%). In terms of income, the distribution is relatively even across different ranges. Car ownership in this cluster is moderate (26%), as is two-wheeler ownership (35%). This cluster can be term as diverse middle class.

Cluster 3 has a higher proportion of females (48%) compared to males (23%). The majority of individuals in this cluster fall into the 18 to 29 years age range (43%). Educationally, this cluster has a higher proportion of individuals with education up to the 10th class (81%). Occupation-wise, homemakers constitute the majority in this cluster (63%). In terms of income, this cluster has a higher proportion of individuals with lower income levels, such as below 10,000 (70%). Car ownership in this cluster is relatively low (10%), and two-wheeler ownership is moderate (24%). They represent non-working population group primarily comprising of students.

Based on the mode usage frequency, it can be seen that Cluster 1 exhibits the highest proportion of users relying on cars as their primary mode of transportation. This cluster, characterized by higher income levels and a dominant business/self-employed occupation, indicates a preference for car ownership and usage. In contrast, both Cluster 2, known as the ‘Diverse Middle-Class’, and Cluster 3, comprised of ‘Young Homemakers’, report much lower reliance on cars. These clusters likely include individuals who either have a more balanced transportation approach or prioritize other modes of travel over private car usage.

Regarding two-wheeler (2W) usage, Cluster 2 emerges as the dominant group, closely followed by Cluster 1. The high 2W ownership in Cluster 2 suggests a preference for this mode of transportation. In contrast, Cluster 3 exhibits the least reliance on two-wheelers, indicating a lower inclination towards using this mode of travel.

Interestingly, the situation is reversed when it comes to Public Transport usage. Cluster 3 reports the highest utilization of public transport. This finding may reflect the demographic composition of Cluster 3, with individuals in this cluster potentially relying on public transport for their daily commuting needs. Cluster 2 follows closely, indicating a significant reliance on public transport as well. In contrast, Cluster 1, which comprises affluent professionals who likely prioritize private transportation options, reports the lowest usage of public transport.

The observations suggest a clear trend between vehicle ownership and public transport usage. Cluster 1, with the highest reliance on cars and comparatively lower usage of public transport, aligns with their preference for private transportation. Conversely, Cluster 3, reporting the least reliance on cars, exhibits the highest usage of public transport. This correlation between vehicle ownership and public transport usage highlights the impact of individual preferences, socioeconomic factors, and lifestyle choices on transportation decisions.

Figure 5. Cluster wise mean score for mode use frequency by purpose.

Coming to usage frequency of smartphone application for trip planning purposes and digital payments as shown in Figure 5, Cluster 3, consisting of young users, demonstrates a higher mean score for app usage related to trip planning across most indicators. This indicates that these individuals rely more heavily on apps for various aspects of their travel planning, such as searching for transportation options, comparing prices, and scheduling trips.

In contrast, Cluster 1 exhibits lower overall app usage, except for the purpose of navigation where it shows the highest use. This higher usage of navigation apps in Cluster 1 can be attributed to their reliance on car transportation, where finding optimal routes and navigating through traffic are crucial. The lower usage of apps for other purposes in Cluster 1 May be influenced by the peculiarities and preferences of relatively older users within this cluster.

Interestingly, Cluster 2 reports the highest mean score for booking cabs, indicating a greater emphasis on using ride-hailing services. This finding aligns with the cluster’s tendency to exhibit multimodal behavior, where individuals are more likely to combine different modes of transportation for their travel needs. Cluster 1 follows closely, showing a relatively high usage of app-based cab booking after navigation. This suggests that even within Cluster 1, there is a significant reliance on app-based cab booking for certain travel requirements.

Overall, these findings highlight the varying patterns of app usage among the different clusters as shown in Figure 6. While Cluster 3, comprising young users, demonstrates a higher reliance on apps for trip planning across multiple indicators, Cluster 1 and Cluster 2 exhibit distinct usage patterns specific to their transportation preferences and needs.

Attitudes and preferences of users: The analysis reveals interesting patterns among different user clusters in relation to their concerns and preferences regarding the digital ecosystem, private vehicle usage, safety, and public transport as shown Figure 7. Cluster 1, composed of older users, demonstrates a higher level of apprehension with the digital ecosystem compared to the other clusters. Surprisingly, cluster 3, consisting of young users, reports even greater concerns than cluster 2, indicating that age does not necessarily correlate with a higher comfort level in the digital realm. The particular aspect where cluster 3 represented by young users report higher degree of concern is the reliability of smartphone application, the reason for both cluster 1 and cluster 3 loading highly on this reason may be different as cluster 1 users may who are generally older may have difficulties dealing with complexities of modern applications whereas young users of cluster 3 May who tend to use and engage these applications more often may not be satisfied with the overall user experience these applications provide.

Cluster 1 stands out for its inclination towards factors that promote private vehicle usage, such as comfort, timesaving, and privacy. These individuals value the convenience and personal space offered by private vehicles. In contrast, cluster 2 exhibits lower scores overall, suggesting a lesser emphasis on these factors. Furthermore, cluster 1 places a greater emphasis on safety, indicating that safety concerns play a prominent role in their decision-making regarding transportation, this could be explained due to higher share of young users and females in this cluster.

Figure 6. Cluster wise mean score for smartphone app usage frequency.

When it comes to public transport, cluster 1 holds a negative perception, consistently reporting higher mean scores across major indicators. It shall be noted that here higher score represents lower inclination towards usage of Public Transport services. Cluster 1 likely perceive public transport as less desirable or inconvenient compared to private vehicles. On the other hand, cluster 2 and cluster 3 view public transport similarly, with lower scores across most important indicators. This suggests that cluster 1 views public transport much more negatively than cluster 2 and 3.

Parking-related issues emerge as a major driver for the use of Ride Hailing services within cluster 1, apart from the factors of ease and perceived low cost. The lack of parking spaces seems to be a significant concern for these individuals, influencing their choice of transportation mode. It shows that parking plays a significant role in the choice between private car and on demand mobility services and can be a controlling factor to limit the use of private vehicles. In contrast, cluster 2 identifies unreliable public transport as a primary reason for utilizing ride-hailing services. This implies that cluster 2 perceives ride-hailing as a more reliable alternative to public transport in certain cases. In contrast, cluster 3 which is found to use Ride Hailing services frequently tend to score lower than other two clusters on all the parameters considered, Specifically, comfort and avoiding negotiations and waiting are the only two parameters where they mean score is above 2.5, which may explain that there are factors other than those considered which result in higher adoption of these services for cluster 3.

Figure 7. Cluster wise mean score for user behavioral attitude data.

The replacement of traditional public transport modes by on-demand mobility services raises important sustainability questions that need to be considered. One of the key concerns is the potential increase in carbon emissions and environmental impact resulting from the shift towards individual ride-hailing services. While ride-hailing services can offer convenience and flexibility to users, it is crucial to evaluate their overall sustainability implications. Another sustainability consideration is the utilization rate of vehicles in ride-hailing services. To reduce private car usage, it is essential to encourage efficient use of available resources. If ride-hailing vehicles are frequently operating with low occupancy rates or traveling long distances without passengers, it can lead to inefficient use of energy and contribute to congestion and emissions. Therefore, promoting measures that encourage shared rides and higher occupancy rates can help improve the sustainability of ride-hailing services. Ride-hailing services can also contribute to reducing private car usage and, consequently, promote sustainability. By providing a convenient and accessible alternative to private car ownership, ride-hailing can encourage individuals to forego owning a car or even consider downsizing their vehicle. This can lead to reduced traffic congestion, lower emissions, and a more efficient use of urban space dedicated to parking.

In cluster 1, the car is the most frequently replaced mode of transportation when it comes to ride-hailing services as shown in Figure 8. This cluster consists primarily of individuals who own private vehicles, and nearly half of them express a willingness to replace their cars with ride-hailing services. This suggests that ride-hailing is seen as a viable alternative to private car ownership for a significant portion of cluster 1. While private vehicles dominate the modes being replaced, there is also a lesser but noticeable impact on public transport within this cluster.

Cluster 2 exhibits a relatively even split among the individual modes of transportation being replaced by ride-hailing services. However, when these modes are combined, they account for more than one third of the potential trips that would have been made by public transport. This implies that cluster 2 users are more open to replacing various modes of transportation with ride-hailing, and as a collective, they contribute significantly to the reduction in public transport usage. Interestingly, among all the clusters, cluster 2 shows the highest number of two-wheeler (2W) and intermediate public transport (IPT) trips being replaced by ride-hailing services.

In contrast, cluster 3, which predominantly relies on public transport as its main mode of travel, is also impacted by the presence of ride-hailing services. While the degree of replacement may not be as substantial as in the other clusters, there are still individuals within cluster 3 who opt for ride-hailing as an alternative to public transport. This suggests that even among users who heavily depend on public transport, there is a segment that considers ride-hailing as a viable option for certain trips or under specific circumstances.

Figure 8. Modes replaced by on demand ride hailing services.

Cluster wise impact on car usage and purchase decision has also been studied. The analysis reveals intriguing insights regarding the impact of ride-hailing services on car usage and purchase decisions across the clusters as shown in Figure 9. Within cluster 1, comprising predominantly car owners, approximately 50% of the users report minimal impact on their car usage. Interestingly, 12% of the users in this cluster do not own a car, leaving roughly 38% who experience some to a significant reduction in their car usage. While clusters 2 and 3 are not primarily dominated by car ownership, they exhibit a similar trend among those who do own cars. Across the clusters, around 40–45% of the users state that their decision to purchase a car remains unaffected by the presence of ride-hailing services. Notably, cluster 1 reports a higher percentage of individuals deferring their car purchase decision compared to the other two clusters. In cluster 3, more than 40% of the users indicate that they have abandoned or given up on their plans to purchase a car altogether.

These findings provide valuable insights into the role of ride-hailing services in influencing car ownership and purchase decisions across different user clusters. The results suggest that while ride-hailing services may not impact everyone’s decision to purchase a car, they do play a significant role in reducing car usage among existing car owners. Moreover, the data highlights the potential of ride-hailing services in dissuading individuals from buying cars, particularly in cluster 3 where a significant percentage of users have opted out of their car purchase plans.

Figure 9. Cluster wise stated impact of ride hailing services on car usage and purchase decisions.

The results obtained from analyzing ride-hailing services offer crucial insights into the potential outcomes of a full-scale introduction of Mobility as a Service (MaaS). The findings highlight the significance of ride-hailing services in mode replacement, particularly in terms of reducing private car ownership and influencing public transport usage. This information is valuable in shaping the development and implementation of a comprehensive MaaS system that integrates various on-demand services to cater to diverse user preferences and promote sustainable transportation practices. By considering these results, decision-makers can design a MaaS framework that optimizes mode choices, enhances user experience, and fosters a seamless and efficient urban mobility ecosystem.

7 Discussions & conclusion

As part of this study, an attempt was made to classify the users into clusters or groups based on their preferences and attitudes towards commute mode, Public Transport, Digital ecosystem & Smartphone app usage and Ride Hailing. Using PCA to reduce the dimensionality and applying K mean clustering technique, the users were classified into 3 distinct clusters by an iterative process to achieve sufficient heterogeneity between the clusters. Differences in Travel pattern, personal vehicle usage, Public Transport usage, App usage for Trip planning, Ride Hailing usage were then studied between the identified clusters to gain a deeper understanding.

The assessment of drivers and barriers in the adoption of Mobility as a Service (MaaS) in an Indian city, as analyzed through the identified clusters, provides significant insights into the key factors influencing the adoption of MaaS. This assessment focused on parameters such as socio-economic and demographic characteristics, private vehicle ownership and usage, public transport usage and attitudes, ride-hailing usage and attitudes, smartphone app usage, and the stated impact on car usage and purchase decisions. By examining these parameters, we can gain a comprehensive understanding of the potential challenges and opportunities in implementing MaaS in the Indian context.

Cluster 1, characterized by a higher proportion of older users with relatively higher income levels, exhibits distinct patterns in their transportation preferences and behaviors. With nearly 90% of users owning at least one car, and a significant portion owning multiple cars, cluster 1 demonstrates a high level of private vehicle ownership and dependence. Their usage of cars is also considerably higher compared to the other clusters, indicating a strong preference for the comfort, convenience, and flexibility offered by private vehicles. However, it is interesting to note that around 50% of cluster 1 users report close to no impact on their car usage, suggesting that ride-hailing services have had limited influence on their transportation choices. On the other hand, a significant proportion of users in this cluster express a willingness to defer the purchase of an additional car, indicating that ride-hailing services could serve as a substitute for additional car ownership among cluster 1 users.

Moving on to public transport usage and attitudes, cluster 1 users exhibit the lowest dependence on public transport among the three clusters. In fact, approximately 70% of users in this cluster reported no use of public transport. Additionally, their attitudes towards public transport are generally negative, as reflected in consistently lower mean scores across major indicators. This suggests that cluster 1 users perceive public transport as less reliable, less convenient, and potentially less comfortable compared to their private vehicles. Therefore, it is important to address their concerns and improve the quality and perception of public transport services to encourage their adoption of MaaS.

Cluster 2, comprising users of all age groups with moderate income levels, shows a more balanced representation in terms of transportation choices. While they also own private vehicles, the level of car ownership is relatively moderate, and a higher proportion of users own two-wheelers. This cluster demonstrates a moderate reliance on private vehicles, but they are more inclined to use public transport compared to cluster 1. Their attitudes towards public transport are not overtly negative, indicating a relatively more favorable perception of the available public transport options. Furthermore, cluster 2 users exhibit a higher usage of ride-hailing services compared to the other clusters, with approximately 20% of users utilizing these services more than 10 times a month. This suggests that ride-hailing services have gained popularity among cluster 2 users, possibly due to factors such as the lack of parking availability and the need for convenient transportation options. However, it is important to note that cluster 2 users report relatively lower scores on the cost of ride-hailing services. This aspect could be addressed through pricing strategies or incentives to encourage their continued use of ride-hailing as a mode of transportation.

Cluster 3, comprising younger users with lower income levels, stands out as the cluster with the highest reliance on public transport. This cluster primarily consists of students, and they utilize public transport for all the purposes asked in the questionnaire, indicating their heavy dependence on this mode of transportation. The mean scores of attitudes towards public transport among cluster 3 users are generally positive, suggesting a relatively favorable perception of the available public transport options. Despite their lower levels of car ownership, cluster 3 users exhibit a moderate to high usage of ride-hailing services. This indicates that ride-hailing services have effectively filled a transportation gap.

7.1 Recommendations

Firstly, cluster 2 and 3 should be prioritized in MaaS implementation efforts. These clusters exhibit moderate reliance on public transport and a higher propensity towards MaaS adoption. To capitalize on this, it is recommended to facilitate the development of user-friendly and innovative platforms and applications that align with the technological inclination of cluster 3 users. These platforms should offer comprehensive trip planning, booking, and payment options, integrating various mobility services. To ensure user satisfaction, active involvement of cluster 3 users in co-creation and feedback processes is crucial.

Furthermore, addressing the specific barriers identified in cluster 1 is essential. This can be achieved through targeted awareness and education campaigns. The campaigns should emphasize the benefits, convenience, and cost-effectiveness of MaaS options. By highlighting the advantages of MaaS, such as reduced reliance on private vehicles, improved access to multiple transportation modes, and reduced congestion and environmental impact, cluster 1 users can be encouraged to consider MaaS as a viable alternative. Additionally, the campaigns should address the concerns raised by this cluster, such as privacy and digital ecosystem acceptance, to alleviate their apprehensions and increase their propensity towards MaaS adoption.

7.2 Future scope of study

It is important to include the influence of emerging 2W Ride Hailing services, such as Rapido, in the assessment. These services have gained popularity and can significantly impact the transportation landscape. Their inclusion will provide a more comprehensive understanding of the evolving mobility options and their implications for MaaS implementation. To ensure a robust assessment, a stratified sampling method can be adopted. This approach will enable the selection of participants from various socio-economic and demographic backgrounds, ensuring a representative sample for each cluster. By capturing diverse perspectives, the assessment can account for variations in preferences and behaviors related to MaaS adoption. To analyze the choices and preferences of each cluster, Stated Preference techniques can be employed to develop choice models. These models will help understand the decision-making processes and trade-offs individuals make when selecting transportation modes. By incorporating the stated preferences of users, valuable insights can be gained on the drivers and barriers in MaaS adoption for each cluster.

Disclosure statement

No potential conflict of interest was reported by the authors.