Article
citation information:
Khursheed, S., Yasmin, S. Diagnostic
evaluation of urban metro transit system post-covid-19. Scientific Journal of Silesian University of
Technology. Series Transport. 2024, 124,
77-91. ISSN: 0209-3324. DOI: https://doi.org/10.20858/sjsutst.2024.124.6.
Salman KHURSHEED[1],
Shagufta YASMIN[2]
DIAGNOSTIC EVALUATION OF
URBAN METRO TRANSIT SYSTEM POST-COVID-19
Summary. Public transportation
usage in Delhi has declined, with the Delhi Metro having a significant share.
However, due to fare hikes and COVID-19 restrictions,
the DM's share has been decreasing further. To improve ridership, a study is
being conducted to evaluate the DM's performance and identify areas for
improvement in passenger convenience and comfort. The Magenta line is
investigated through an on-board survey to collect primary data. The survey
covers commuter perceptions of safety & security, financial & economic
factors, infrastructure & comfort and functional & operational
features. The Relative Importance Index approach is used to analyse the data
and evaluate DM performance. An ANN model is also presented to determine the
factors influencing the choice to travel on the DM, with the “metro fare per
trip” factor being a key consideration. Based on the analysis results,
recommendations are made to improve the DM's performance. The study found that
safety and security had the highest RII, followed by
efficiency and viability, functional and operational features, infrastructure
and comfort, and financial and economic factors. The subway fare had the lowest
RII. The ANN model is adapted to understand the
reasons behind low metro ridership.
Keywords: Delhi Metro, Post-Covid-19,
Public Transportation, Performance Evaluation, Performance Indicators
1. INTRODUCTION
The Delhi Metro (DM) is one of the world's largest and extensive metro
networks. The DM is a major means of Public Transportation (PT), making commutes
accessible to millions. The DM has a distance-based fare system and
infrastructure with future network expansion (FE Online, 2017). The DM saw
decreased ridership due to an increase in fare (Sanjana
Agnihotri, 2018). To mitigate the influence of COVID-19, the DM set limits and social distancing tactics.
The DM's ridership was curtailed due to the travel guidelines during COVID-19 (Jasjeev Gandhiok, 2022). Even when the limitations were lifted, the
DM suffered losses due to decreased ridership (Atul Mathur, 2023).
The inability to fully utilise DM as a PT and the lack of Last Mile
Connectivity (LMC) are other factors in the drop in
usage (Mittal, 2023). The author suggests that the attractiveness of PT relies
on safety, cost, time, comfort, and convenience. Transit planning often
neglects first and last-mile connectivity (Board et al., 1973). The rationale
for the fall in ridership of DM must be determined. This can be comprehended
using the Performance Evaluation (PE) of DM (Khursheed
& Kidwai, 2022). The Magenta (Line 8), of DM has
a total length of 37.46 km and consisting of 25 metro stations from Janakpuri West to the Botanical Garden (DMRC,
2022). A performance evaluation on the magenta line is essential given its
length and the decline in ridership in DM.
The paper explores a comprehensive analysis of various methodologies used
to assess the performance of the Magenta Line and provides a thorough
investigation of the multifaceted issues related to ridership, providing
valuable insights and recommendations for future enhancements.
2. LITERATURE REVIEW
A robust, inexpensive, and safe PT infrastructure is crucial. It is to
ensure everyone has equal access to opportunities in work, education,
healthcare, and recreation, especially for those who are socially and
economically disadvantaged. The PT is a vital facility that must be provided
equitably (Ghosh et al., 2022). Currently, the majority of riders are
middle-class office workers and students of all age groups, rather than the
fair distribution of riders that the DM had envisioned. The cause for this is
an increase in metro fares in quick succession that is no longer affordable to
the poor/middle class of the society (Saraswat &
Girish, 2020). The PT provided by the metro has a high possibility of usage if
it is easily accessible by the users in terms of time, distance, safety, and
convenience (Bhandari et al., 2014). Long total travel time and high cost
lowers the probability of using the PT. The increase in access-egress distance
to the transfer location reduces the chances of using the PT (Keijer and Rietveld 2004; Loutzenheiser
1997; O’Sullivan and Morral 1996). The metro projects
in India are implemented in isolation without concern for access and egress
connectivity (Geetam Tiwari, 2013). There is growing
recognition of the importance of LMC to mass transit
systems. The authors recommend that walking as an LMC
choice needs to be promoted through enhanced user experience, in the absence of
which a significant amount of last-mile travel will happen through
unsustainable mechanized modes. (Kanuri et al., 2019). Service quality
characteristics, notably safety, and security, have the greatest influence on
passenger adoption behaviour, favourably impacting DM as a PT. (Yogendra Pal Bharadwaj, Mukesh Singh, 2020). The methods for evaluation of a
transport project now draw indicators accounting for accessibility, safety, and
environmental effects (Lake and Ferrari 2002, Zhu and Ziu
2004). Factors such as fares, quality of service, and ownership of vehicles also
influence the use of PT (Neil Paulley.et.al, 2006).
Before COVID-19, these investigations were
carried out by researchers from various regions and timelines. The PIs from the
preceding macro-level studies are incorporated in the present DM micro-study.
As a result, based on the scope of the research indicated above, this post-COVID micro-level study is undertaken on the MAGENTA line.
The study considers infrastructure and comfort, security and safety, functional
and operational, efficiency and viability, and financial PID
to assess the performance of the MAGENTA line. An Artificial Neural Network
(ANN)-based performance model considering private car ownership, major haul
distance, DM parking facilities, frequency of DM travel, and number of station
interchanges per trip. The time for access and egress is proposed. The model's
goal is to determine the PI factors influencing the Magenta line performance.
3. RESEARCH METHODOLOGY
Understanding
commuter preferences within the DM demands a nuanced approach that combines the
use of both qualitative and quantitative approaches. This study involves an
in-depth analysis of multifaceted factors that influence commuters' mode
choices. To facilitate these strategies, a significant number of passengers'
travel data is required. A preliminary survey was conducted among a small
cohort to refine and enhance the survey questionnaire before a wider
distribution. Feedback from this preliminary phase is used to redesign the
questionnaire and make it more holistic, to encompass the operational,
financial, safety and comfort considerations within the DM framework.
The
parameters were examined, and a logical analysis was applied to ensure the
reliability and accuracy of the data used. The check was conducted on board
with an aggregate of 742 passengers from which 630 passengers' answers were
anatomized for keeping in mind the considerable size to understand the commuter
preferences comprehensively. Post-data collection, a statistical interpretation
was done leveraging the various software. This method facilitates a structured
organization of extensive questionnaire data, providing an overview of the
commuter inclinations.
The
cross-tabulation analysis is an important statistical tool in enabling the
interpretation of several factors, focusing on PIs and the RII.
The PI serves a critical role in channelling the
variables into distinct channels while encompassing the nuanced perspectives of
commuters. Simultaneously, the RII helps in ranking
the variables and identifying the factors influencing the metro ridership. The
ANN model facilitates the decision-making process by unravelling intricate
patterns within these complex data structures. This analysis offers a
comprehensive evaluation of the indices dictating the commuter preferences
within the DM framework.
The
subsequent sections of this research paper provide insights into the analysis
and interpretation of the implications drawn from survey outcomes. This
provides a base for the decision-making process in the ridership analysis of
DM.
4. DESCRIPTIVE AND CROSS-TABULATION
ANALYSIS
The
Descriptive analysis serves as a precise instrument to methodically organize
data, presenting a structured outline of its elements and variables.
Conversely, the cross-tabulation analysis details the interdependencies within
the dataset components (Widyaningsih. et, al., 2022). The development of transportation networks
involves a multitude of factors that influence their utilization and
flexibility. It involves demographic aspects such as age, frequency,
accessibility, parking provisions, comfort, ticket convenience, and amenities.
Tab. 1
Travel frequency vs age of commuters
|
Age of Commuters (Years) |
Occasional (%) |
1-2 times per month (%) |
4-5 times per month (%) |
2-3 times per month (%) |
Daily (%) |
Total (%) |
|
Less than 20 |
4 (5.9) |
2 (8.4) |
5 (10) |
8 (9) |
14 (7.2) |
33 (7.7) |
|
20-30 |
32(47.9) |
14 (58.3) |
31 (62) |
42(47.2) |
107(55.2) |
77(18.1) |
|
30-40 |
20 (29.4) |
4 (16.7) |
7(14.0) |
21(23.6) |
48 (24.7) |
100 (23.5) |
|
40-50 |
6 (8.8) |
4 (16.7) |
5 (10) |
13 (14.6) |
17 (8.8) |
45 (10.6) |
|
50-60 |
5 (7.4) |
0 (0) |
2 (4) |
5 (5.6) |
5 (2.6) |
17 (4) |
|
Above 60 |
1 (1.5) |
0 (0) |
0 (0) |
0 (0) |
3 (1.5) |
4 (0.9) |
|
Total |
68 (16) |
24 (5.6) |
50 (11.8) |
89 (20.9) |
194 (45.6) |
425 (100) |
Source:
[13, 14]
The survey
conducted was analysed, and the strategies were
applied to explore the correlation between commuters' age and the frequency of
utilizing the metro. It is noted that the age group of 20–30 years constitutes
a large segment of the daily metro users, followed by the 30-40 age group. It
is also observed that the age group above 40 has the lowest number of users.
This underscores that DM usage is most prominent among the 30-50 age group who
travel using the metro for educational and occupational purposes. The age group
above 50 does not prefer the metro. These findings reveal the imperative need
to improve the DM services to ensure inclusivity across different age cohorts
and to address concerns.
5. PERCEPTION INDEX BASED ON PERFORMANCE
INDICATORS (PID)
The PI
assesses the transit performance to ensure a continuous increase in the quality
of the transit services and to allocate resources using PIDs
(Khursheed & Kidwai,
2023). Evaluating transit service quality involves both subjective measures of
passengers' perceptions and metrics. They are compared to gauge the quality of
transit services and detect opportunities for enhancement. (Hounsell,
Mathew, 2023). The PID from the European standards
are considered to interpret the survey and understand the limitations. The PT
quality determinants have been studied extensively, and services are mainly
characterized by several aspects such as service availability, reliability,
comfort, cleanliness, safety and security, fare, information, and customer
care. These aspects can be measured in various ways by considering different
indicators (Eboli, L. and Mazzulla,
G., 2012). The following subsections will provide a detailed description of some
of these indicators, along with suggested target values.
The PI is
composed of multiple evaluations that assess different aspects of the
transportation network. These evaluations include the fairness of metro fares
compared to other modes of public transport, satisfaction with station parking
facilities, frequency of the DM, overall satisfaction with DM services, and the
effectiveness of nighttime security measures. A study
has shown that the total duration of DM trips compared to other transportation
options, as well as the higher expenses associated with the metro, has been
correlated with a significant number of dissatisfied PI ratings. Many DM
customers, especially those travelling shorter distances, are requesting fare
adjustments to align them with other modes of transportation. Furthermore, the
data suggests that there is room for improvement in various aspects of the PI
ratings, particularly regarding the LMC.
Table 2
examines the perceptions regarding different performance indicators that impact
DM usage. These evaluations were conducted to gauge the quality or
effectiveness of each specific indicator in influencing usage patterns. This
analysis aims to understand how various factors contribute to the overall
assessment of DM, providing insights into which indicators are deemed
significant in shaping users' decisions to utilize this mode of transportation.
The quality assessment of these indicators helps in comprehending their
respective impacts on the usage patterns of DM services.
Tab. 2
Perception index based on performance indicators
|
S.
No. |
Performance parameters |
Yes |
No |
Perception Index |
|
A |
Infrastructure and Comfort Performance Indicators |
|
|
|
|
1 |
Are parking facilities offered by Delhi Metro enough
and affordable? |
69.17% |
30.83% |
6.29 |
|
2 |
Are there sufficient vending machines and easy to
use? |
81.70% |
18.30% |
|
|
3 |
Is there sufficient standing space for passengers? |
61.54% |
38.46% |
|
|
4 |
Is there sufficient seating space for passengers? |
39.25% |
60.75% |
|
|
B |
Security and Safety Performance Indicators |
|
|
|
|
1 |
Do you think Security measures
like CCTV cameras at stations and metro coaches are effective for safety at
night? |
96.28% |
3.72% |
8.80 |
|
2 |
Do you think that frisking and
X-ray checking of luggage's at stations are
effective for security measures? |
91.44% |
8.56% |
|
|
3 |
Do you think metro train services
are required after 11:30 PM as well? |
76.35% |
23.65% |
|
|
C |
Functional and Operational Performance Indicators |
|
|
|
|
1 |
Are you a frequent traveller by metro? |
60.77% |
39.23% |
5.77 |
|
2 |
Are you satisfied with the
operating frequency of Delhi Metro services at office hours? |
65.58% |
34.42% |
|
|
3 |
Does travelling by metro
increases your time? |
83.91% |
16.09% |
|
|
4 |
Do you think breakdowns in Metro
cause hindrance in your working routine? |
22.73% |
77.27% |
|
|
5 |
Do you consider other
transportation means due to delays and breakdown of Metro trains in your
daily working routine? |
28.44% |
71.56% |
|
|
6 |
Does it create any trouble while
using interchange? |
84.78% |
15.22% |
|
|
D |
Financial and Economic Performance Indicators |
|
|
|
|
1 |
Do you think that Metro fares are
costlier comparing other public transport systems in Delhi? |
63.16% |
36.84% |
2.76 |
|
2 |
Do you think that there should be
reduction in fare in metro? |
81.70% |
18.30% |
|
|
E |
Efficiency and Viability Performance Indicators |
|
|
|
|
1 |
Are you satisfied with the
services provided by Delhi Metro Rail Corporation? |
90.16% |
9.84% |
6.21 |
|
2 |
Do you think Delhi Metro Rail Corporation is
efficient? |
81.14% |
18.86% |
|
|
3 |
Do you think there is a need to improve Last Mile Connectivity? |
41.03% |
58.97% |
|
|
4 |
Do you think there is a need to increase the number
of trains? |
35.20% |
64.80% |
|
|
5 |
Do you find metro network simple? |
63.19% |
36.81% |
Source: Authors
The data
suggests that safety and security are paramount to DM users, as reflected in
the highest rating given to these aspects. Additionally, the survey highlights
a willingness among commuters to utilize the metro service beyond 11:30 pm,
indicating a demand for extended operational hours. The survey revealed that
financial considerations, particularly fare costs, received the least
attention. This insight implies that fare pricing might significantly influence
commuters' choices, leading them to opt for alternative transportation modes
over the metro.
In essence,
the priority given to safety and the desire for extended service hours
demonstrates the positive aspects of the metro. However, addressing fare
concerns could be pivotal in retaining and attracting more commuters to the
metro system.
6. RELATIVE IMPORTANCE INDEX OF PERFORMANCE INDICATORS
The RII serves as a useful means of assessing metro performance
indicators. It assigns a numerical value to the importance or relevance of
various indicators, enabling the prioritization of elements based on their
perceived significance in the realm of DM services. Through the analysis of
survey data or opinions about metro performance indicators, the RII provides valuable insights into the factors that hold
the greatest sway over overall metro performance, facilitating targeted enhancements
and strategic decision-making within metro systems (Kurniawati,
et.al. 2023).
(1)
Here RII is the relative importance index, W is the weighting
assigned to each element by respondents (ranging from 1 to 5), A is the
greatest weight (in this example, 5), and N is the total number of respondents.
The RII value ranges from 0 to 1 (0 inclusive); the
greater the RII, the more significant the indication.
Table 3 displays the RIIs of PID
as well as the findings.
Understanding
the intricacies and nuances of PID within transit
systems offers a profound glimpse into the facets shaping passenger experience.
The evaluation, rooted in user feedback and satisfaction rankings, unveils both
commendable aspects and critical areas necessitating attention. At the pinnacle
of satisfaction rankings lies the Security and Safety PID,
a testament to meeting and potentially surpassing passenger expectations. This
vital aspect, foundational to trust and comfort, stands as an exemplar within
transit systems. Following closely are the Efficiency & Viability indices,
marking significant contributions to positive passenger experiences. However, RII exposes focal points for improvement. Concerns over LMCs, limited seating space within coaches, and inadequate
designated parking facilities emerge as pivotal areas warranting attention.
Addressing these concerns is crucial for enhancing overall passenger
experiences and boosting ridership.
A striking
disparity surfaces when comparing the highest-rated DM efficiency PID against the lowest-rated metro fare in terms of RII and satisfaction rank. This incongruity highlights the
impact of fare structures on passenger contentment and ridership trends,
particularly in the context of the COVID-19 epidemic.
The correlation underscores the need for a nuanced reassessment of fare
strategies that prioritize accessibility and passenger satisfaction while
aligning with economic imperatives. LMC assumes
paramount importance in transit systems. Optimizing this aspect through
seamless integration and accessibility can significantly elevate the overall
transit experience, addressing concerns and fostering inclusivity and
efficiency. The limitation of seating space within coaches poses a tangible
challenge. Striking a balance between capacity and comfort becomes imperative
to accommodate increasing passenger demands. Innovative design interventions or
operational strategies can optimize space utilization without compromising
passenger comfort. In conclusion, the comprehensive evaluation of PID within transit systems outlines a transformative
trajectory. Strategic interventions in fare structures, last-mile connectivity,
seating space optimization and parking facilities are pivotal in sculpting
transit systems that prioritize passenger satisfaction and accessibility. This
holistic reimagining, underscored by innovation and collaboration, fosters
systems that harmonize efficacy, accessibility, and passenger-centricity. Table
3 shows the ranking index from each indicator and gives an overall rank for the
subcategories.
Table 3
assesses perceptions concerning various PIs influencing DM usage. The
evaluations aim to determine the quality or effectiveness of each indicator in
influencing the patterns of usage. The indicators are ranked from 1 to 19,
reflecting their importance in terms of performance. Within the infrastructure
and comfort category, the ease of using vending machines received the highest
rank, while seating spaces were rated the lowest. This suggests that commuters
found vending machine accessibility more satisfactory compared to seating
availability. In terms of safety and security, the overall index received the
highest ranking, indicating that users highly prioritize safety measures.
However, the dissatisfaction with the availability of metro services after
11:30 pm resulted in this specific sub-indicator receiving the lowest ranking
within this category.
Tab. 3
RII
ranking of performance indicators based on users' perception
|
S. No. |
Performance parameters |
RII |
Overall RII |
Rank |
Overall Rank |
|
A |
Infrastructure and Comfort Performance Indicators |
|
|
|
|
|
1 |
Are parking facilities offered by Delhi Metro enough and affordable? |
0.605 |
0.609 |
3 |
13 |
|
2 |
Are there sufficient vending machines and easy to use |
0.695 |
1 |
8 |
|
|
3 |
Is there sufficient standing space for passengers? |
0.650 |
2 |
12 |
|
|
4 |
Is there sufficient seating space for passengers |
0.486 |
4 |
19 |
|
|
B |
Security and Safety Performance Indicators |
|
|||
|
1 |
Do you think security measures
like CCTV cameras at stations and metro coaches are effective for safety at
night? |
0.855 |
0.792 |
1 |
1 |
|
2 |
Do you think that frisking and
X-ray checking of luggage's at stations are
effective for security measures? |
0.850 |
2 |
2 |
|
|
3 |
Do you think metro train services
are required after 11:30 PM as well? |
0.673 |
3 |
10 |
|
|
C |
Functional and Operational Performance Indicators |
|
|||
|
1 |
Are you a frequent traveller by metro? |
0.591 |
0.652 |
4 |
14 |
|
2 |
Are you satisfied with the
operating frequency of Delhi Metro services at office hours? |
0.700 |
3 |
7 |
|
|
3 |
Does travelling by metro increase
your time? |
0.791 |
2 |
6 |
|
|
4 |
Do you think a breakdown in the
Metro cause hindrance in your working routine? |
0.500 |
5 |
17 |
|
|
5 |
Do you consider other
transportation means due to delays and breakdown of the Metro trains in your
daily working routine? |
0.495 |
6 |
18 |
|
|
6 |
Does it create any trouble while
using interchanges? |
0.836 |
1 |
3 |
|
|
D |
Financial and Economic Performance Indicators |
|
|||
|
1 |
Do you think that the Metro fares
are costlier than other Delhi public transport systems? |
0.691 |
0.693 |
2 |
9 |
|
2 |
Do you think that there should be
a reduction in fares in the metro? |
0.695 |
1 |
8 |
|
|
E |
Efficiency and Viability Performance Indicators |
|
|||
|
1 |
Are you satisfied with the
services provided by the Delhi Metro Rail Corporation? |
0.832 |
0.676 |
1 |
4 |
|
2 |
Do you think the Delhi Metro Rail Corporation is efficient? |
0.795 |
2 |
5 |
|
|
3 |
Do you think there is a need to improve Last Mile Connectivity? |
0.532 |
5 |
16 |
|
|
4 |
Do you think there is a need to increase the number of trains? |
0.568 |
4 |
15 |
|
|
5 |
Do you find the metro network simple? |
0.655 |
3 |
11 |
|
Source:
Authors
The
indicator with the lowest overall RII was seating
space, mainly due to its limited availability. This suggests that the provision
of adequate seating is a crucial factor affecting commuter satisfaction and
usage patterns. This analysis aims to understand how various factors contribute
to the overall assessment of DM, providing insights into which indicators are
deemed significant in shaping users' decisions to utilize this mode of
transportation. The quality assessment of these indicators helps in
comprehending their respective impacts on the usage patterns of DM services.
7.
PERFORMANCE EVALUATION MODEL USING ARTIFICIAL NEURAL ANALYSIS (ANN)
Artificial
Neural Networks (ANNs) are widely recognized as
valuable and resilient computational models for both prediction and
classification tasks. To enhance the accuracy and reliability of the processed
data, these architectures are frequently constructed using a well-suited
amalgamation of artificial neurons and activation functions (Nandal et al., 2020). ANNs,
specifically Multilayer Perceptron (MLP) models, are
highly effective tools in modern transportation research. They offer a
sophisticated approach to evaluating performance indicators and measuring
passenger satisfaction in transit systems. The implementation of an MLP-based ANN model involves a complex network of
interconnected nodes that enables the system to learn from input data,
recognize patterns, and make predictions (Qu & Chen, 2008). In the context
of the DM, utilizing an MLP-based ANN model is
crucial as it allows for a comprehensive analysis of various performance
metrics, including travel time, service frequency, and passenger sentiments.
This approach provides a holistic understanding of the system's functionality.
The
architecture of the MLP-based ANN model for
evaluating DM performance incorporates a range of input variables. These
variables encompass factors such as travel duration, distance between stations,
train frequency, service reliability, station amenities, and passenger
feedback. Through meticulous data collection and preprocessing,
the model undergoes multiple training iterations to optimize its parameters.
This process enhances its ability to identify complex relationships between
these variables and predict performance indicators. One of the model's key
features is its ability to forecast passenger satisfaction levels based on
collected data. By assimilating feedback on station facilities, journey
experiences, and perceived service quality, the MLP-based
ANN model provides valuable insights into commuters' sentiments. It analyses
their preferences and highlights areas that require attention within the DM
(Gallo et al., 2019). This predictive capability serves as a guide for transit
authorities, enabling them to take proactive measures to address shortcomings
and enhance the overall passenger experience while aligning with their needs
and expectations.
Tab. 4
SSE and RMSE values
|
Training |
Testing |
||||
|
SS |
SSE |
RMSE |
SS |
SSE |
RMSE |
|
307 |
12.250 |
0.199 |
118 |
5.087 |
0.207 |
|
288 |
11.19 |
0.197 |
131 |
4.627 |
0.187 |
|
283 |
10.789 |
0.195 |
142 |
6.574 |
0.215 |
|
292 |
9.654 |
0.181 |
133 |
5.454 |
0.202 |
|
306 |
10.184 |
0.182 |
119 |
4.214 |
0.188 |
|
293 |
11.594 |
0.198 |
132 |
4.403 |
0.182 |
|
298 |
11.867 |
0.199 |
127 |
4.217 |
0.182 |
|
300 |
11.404 |
0.195 |
125 |
5.658 |
0.212 |
|
309 |
10.627 |
0.185 |
116 |
5.884 |
0.225 |
|
292 |
10.262 |
0.187 |
133 |
5.602 |
0.205 |
|
Mean |
10.982 |
0.192 |
Mean |
5.172 |
0.200 |
|
SD |
0.821 |
0.0072143 |
SD |
0.795 |
0.014960 |
Source:
Authors
Fig.
1. RMSE training and testing
The MLP structure consists of interconnected nodes organized
into layers, encompassing input, hidden, and output layers. The design of the
network involves configuring the number of neurons in each layer, defining
activation functions, and establishing connections. The ANN model is trained
using a dataset split into training and testing subsets (Gallo et al., 2019).
During training, the network iteratively adjusts weights and biases to minimize
errors, optimizing its ability to predict outcomes. Post-training, the model's
performance is assessed by computing the Sum of Squared Errors (SSE) and
subsequently deriving the Root Mean Square Error (RMSE)
of given sample size (SS). These metrics quantify the variance between
predicted and actual values, determining the model's predictive accuracy.
The RMSE calculations serve as pivotal measures in model
evaluation. The RMSE computed on training data gauges
the model's fit to the data it was trained on, while the RMSE
on testing data assesses the model's ability to generalize to new, unseen data.
Lower RMSE values on both training and testing sets
indicate superior model performance and stronger generalization capabilities.
Additionally, sensitivity analysis is conducted to scrutinize the model's
robustness. By systematically varying input variables and observing resultant
changes in model output, this analysis elucidates how modifications in inputs
influence the model's predictions, offering insights into the model's
sensitivity to alterations in specific variables (Buran & Erçek, 2022).
These
methodologies collectively contribute to comprehensively assessing and
enhancing the ANN with MLP model's performance and
reliability. Through training, RMSE computations, and
sensitivity analyses, the model's predictive accuracy, generalization capacity,
and sensitivity to input variations are systematically evaluated and refined.
These processes are fundamental in iteratively improving the model's efficacy,
ensuring its applicability in predicting outcomes and informing decision-making
processes within the context of the study (Nurhadi et
al., 2014).
In
sensitivity analysis for each iteration, the normalized significance (NI) of
the various neurons is calculated and presented as a percentage of influence in
that iteration. The NI is defined as the percentage ratio of each factor's
significance over the maximum importance. Furthermore, the normalized value
(NV) of each neuron is derived by dividing its average of normalized importance
(AvNI) by its maximal significance and is shown as a
ratio (Naser et al., 2020).
Tab. 5
Sensitivity analysis
|
IV |
NI (1) |
NI (2) |
NI (3) |
NI (4) |
NI (5) |
NI (6) |
NI (7) |
NI (8) |
NI (9) |
NI (10) |
AvNI |
NV |
Rank |
|
G |
10.7 |
9.1 |
2.5 |
6.5 |
5.5 |
8.3 |
5 |
12.9 |
18.9 |
9.7 |
0.085 |
0.091 |
6 |
|
OV |
14.0 |
16.7 |
18.4 |
33.1 |
9.2 |
18.3 |
10 |
16.7 |
33.4 |
17.6 |
0.176 |
0.188 |
5 |
|
PF |
27.7 |
19.3 |
18.6 |
13.1 |
9.9 |
16.8 |
27.2 |
13.8 |
34.8 |
50.8 |
0.222 |
0.238 |
4 |
|
NS |
31.0 |
46.2 |
47.4 |
31.2 |
49.4 |
56.7 |
39.4 |
34.6 |
100 |
48.3 |
0.453 |
0.485 |
2 |
|
(A+E) Cost |
100 |
100 |
100 |
100 |
100 |
100 |
100 |
100 |
80.3 |
100 |
0.932 |
1.000 |
1 |
|
Age |
38.1 |
20.7 |
22.3 |
18.8 |
16.6 |
12.3 |
10% |
35.7 |
43.1 |
14.8 |
0.226 |
0.242 |
3 |
Source:
Authors
In the above
table, the dependent variable is considered to be Metro fare per day and the IV
= independent variable is G=Gender, OV= Ownership of
Vehicle, PF=Parking facility, NS= No. of station interchanges, (A+E) Cost= Access and Egress cost and Age of commuters. The
NI are presented in percentage share of the PIs in the respect iterations.
From Table 5
we can understand that the (A+E) Cost, which is the
access and egress cost, is the determinant factor. The access and egress cost
corresponds to the distance between the metro and the access-egress locations
and the cost incurred to cover the same. It involves the last mile
connectivity, which governs the access and egress time. The number of station
interchanges has a significant impact on the DM ridership. The age as discussed
in Table 1 determines the ridership due to factors such as health and other
issues that are related to age. On the other hand, ownership of vehicles,
parking facilities and gender have the lesser impact on the metro ridership.
These findings are in confirmation to the explanation noted in another research
authored by (Khursheed & Kidwai,
2022).
8. CONCLUSION AND RECOMMENDATIONS
The age groups
of 20-30 years and 30-40 years are the active riders of DM and their concerns
are needed to be addressed on priority. The 20-30 year age group contributes to
50% of the ridership every day. The RII gives a
ranking based on the perception of metro riders. Safety and security have the
highest ranking but the infrastructure and comfort stand at the lowest rank due
to less seating space available in the metro. These observations are similar to
those (Khursheed & Kidwai,
2022). From the ANN model, it is noted that the access and egress costs are
noted to be the most dominating factor. It is inferred that access-egress trip
fare plays a major role in metro ridership. From the studies and research, it
is concluded that the economic and comfort factors need to be
improved/rationalized to have increased metro ridership in the Magenta Line.
9. LIMITATIONS AND FUTURE SCOPE OF WORK
The study is
confined to on-board passengers, and subsequent research would require a survey
of non-metro commuters. The poll might be affected by how people commute and
what they prefer in different weather conditions. As a future study focus, the
other PID in terms of financial and economic,
functional and operational elements may be investigated. The impact of the
aforementioned elements on other cities in the country must be investigated.
References
1.
Bhandari
K., M. Advani, P. Parida, S. Gangopadhyay. 2014. „Consideration of access and egress
trips in carbon footprint estimation of public transport trips: case study of
Delhi”. Journal of Cleaner Production 85: 234-240. DOI: 10.1016/j.jclepro.2014.05.013.
2.
Stopher P.R., B.D. Spear, P.O. Sucher. 1974. „Towards the Development of Measures
of Convenience for Travel Modes”. Transportation Research Record 527: 16-32.
3.
Buran
B., M. Erçek. 2022. „Public transportation business model
evaluation with Spherical and Intuitionistic Fuzzy AHP and sensitivity analysis”. Expert
Systems With Applications 204: 1-8. DOI: 10.1016/j.eswa.2022.117519.
4.
Delhi
Metro Magenta Line Route Map. Available at: https://www.delhimetrotimes.in/delhi/maps/delhi-metro-magenta-line.html.
5. Eboli L., G. Mazzulla. 2012. „Performance indicators for an
objective measure of public transport service quality”. European Transport \ Trasporti
Europei 51:
1-4.
6.
Delhi
Metro fare hike chart: DMRC raises ticket prices, take a look at full list of
revised fare. Financial Express. 2017.
Available at: https://www.financialexpress.com/india-news/delhi-metro-fare-hike-chart-dmrc-raises-ticket-prices-take-a-look-at-full-list-of-revised-fare/888315/.
7.
Gallo M., G. De Luca, L. D’Acierno, M. Botte. 2019. “Artificial Neural Networks for
Forecasting Passenger Flows on Metro Lines”. Sensors 19(15): 3424. DOI: 10.3390/s19153424.
8.
Gandhiok
J. 2022. „Ridership yet to
hit pre-pandemic levels: Metro data. Hindustan Times”. Available at: https://www.hindustantimes.com/cities/delhi-news/ridership-yet-to-hit-pre-pandemic-levels-metro-data-101650843276426.html.
9.
Ghosh T., T. Kanitkar, R. Srikanth. 2022. „Way Forward to Provide Affordable, Accessible, and Sustainable
Transport for Megacities”. Social Science Research Network: 1-30. DOI: 10.2139/ssrn.4211070.
10. Hounsell M. 2023. „Empirical analysis supporting
research to develop Collaborative Multi-Modal Customer-Centric Key Performance
Indicators for Transport”.
DOI: 10.13140/RG.2.2.17438.51520/1.
11. Kanuri C., K. Venkat, S. Maiti, P. Mulukutla.
2019. „Leveraging
innovation for last-mile connectivity to mass transit”. Transportation
Research Procedia 41: 655-669. DOI: 10.1016/j.trpro.2019.09.114
12. Keijer M.J.N., P. Rietveld. 2004. „How do people get to the railway
station; a spatial analysis of the first and the last part of multimodal trips”. Journal
of Transport Planning and Technology 11(3): 265-275.
13. Khursheed S., F.A. Kidwai. 2021. “Post COVID Performance Evaluation of Delhi Metro”. In: Proceedings of the 2nd
International Conference on Futuristic and Sustainable Aspects in Engineering
and Technology (FSAET-21). Mathura, India. December 24-26, 2021. DOI: 10.1063/5.0154102.
14. Khursheed S., F.A. Kidwai. 2022. “Post COVID-19 Access Egress
Attributes for Urban Metro Transit Users in Delhi”. Journal of Applied
Engineering Sciences
12(25): 53-60.
DOI: 10.2478/jaes-2022-0009.
15. Khursheed S., F.A. Kidwai. 2022. “Evaluation of Users Approbation Indicators
of Delhi Metro”. In: Proceedings of the 8th International
Conference on Transportation System Engineering & Management. NIT-Calicut and selected for
publication in book series “Recent Advances in Transportation Systems
Engineering and Management”. P. 349-365. DOI: 10.1007/978-981-19-2273-2_24.
16. Khursheed S., F. Ahmad Kidwai. 2022. “Post-COVID-19 Performance Evaluation of Urban Metro Transit System in
Delhi and Influence on Access Mode”, Case
Studies on Transport Policy 10(3): 1862-1871. DOI: 10.1016/j.cstp.2022.07.015.
17. Kurniawati Y.R., J. Sumabrata, G.N. Christin. 2023. „Analysis Of
The Variables Affecting Krl Commuter Line Passengers' Choices on The Feeder
Modes To And From Bogor Station”. International Journal Of
Engineering Advanced Research 5(2): 1-10.
18. Lake M., L. Ferreira. 2002. Towards A Methodology To Evaluate Public
Transport Projects. Available at: https://www.researchgate.net/publication/27464625_Towards_A_Methodology_To_Eval.uate_Public_Transport_Projects/link/5417527f0cf2218008bee35e/download
19. Liu S., X. Zhu. 2004. „An Integrated GIS Approach to
Accessibility Analysis”.
Transactions in GIS 8: 45-62. DOI: 10.1111/j.1467-9671.2004.00167.x.
20. Loutzenheiser D.R. 1997. „Pedestrian access to transit: model
of walk trips and their design and urban form determinants around Bay Area
rapid transit stations”.
Transportation Research Record 1604: 40-49.
21. Mathur A. 2023. „Yet to fully recover from the
pandemic blow, Delhi Metro is in red for the second year”. The
Times of India. Available at: https://timesofindia.indiatimes.com/city/delhi/yet-to-fully-recover-from-pandemic-blow-metro-in-red-for-secondyear/articleshow/97502612.cms.
22. Mittal T. 2023. „New Delhi metro must keep up as
capital becomes world’s largest city by 2028”. ThePrint.
Available at: https://theprint.in/india/new-delhi-metro-must-keep-up-as-capital-becomes-worlds-largest-city-by-2028/1681173/.
23. Nandal M., N. Mor, H. Sood. 2020. „An Overview of Use of Artificial
Neural Network in Sustainable Transport System”. Advances
in Intelligent Systems and Computing 1227: 83-91. DOI: 10.1007/978-981-15-6876-3_7.
24. Naser I.H., A.M. Mahdi, Y.H. Jasim. 2020. “Performance of Artificial Neural
Networks (ANN) At Transportation Planning Model”. IOP
Conference Series: Materials Science and Engineering 928: 022-032. DOI: 10.1088/1757-899x/928/2/022032.
25. Nurhadi L., S. Borén, H. Ny. 2014. „A Sensitivity Analysis of Total Cost
of Ownership for Electric Public Bus Transport Systems in Swedish Medium-Sized
Cities”. Transportation Research Procedia 3: 818-827. DOI: 10.1016/j.trpro.2014.10.058
26. O’Sullivan S., J. Morrall. 1996. „Walking distances to and from
light-rail transit stations”. Transportation Research Record
1538: 19-26. DOI: 10.1177/0361198196153800103.
27. Paulley N., R.J. Balcombe, R. Mackett,
H. Titheridge, J. Preston, M. Wardman, J. Shires, P. White. 2006. „The demand for public transport: The
effects of fares, quality of service, income and car ownership”. Transport
Policy 13(4): 295-306. DOI: 10.1016/j.tranpol.2005.12.004.
28. Qu L., Y. Chen. 2008. „A Hybrid MCDM Method for Route
Selection of Multimodal Transportation Network. Lecture Notes in Computer Science: 374-383. DOI: 10.1007/978-3-540-87732-5_42.
29. Sanjana Agnihotri. 2018. „Delhi Metro daily ridership 32 per
cent less than projected in 2018: CSE”. India Today. Available
at: https://www.indiatoday.in/india/story/delhi-metro-daily-ridership-32-per-cent-less-than-projected-in-2018-cse-1332172-2018-09-05.
30. Saraswat I., K. Girish. 2020. „The Delhi Metro: A rail of issues”. Available at: https://www.researchgate.net/publication/346646804_The_Delhi_Metro_A_rail_of_issues.
31. Tiwari G. 2013. „Metro Rail and the City: Derailing
Public Transport”.
Economic and Political Weekly 48(48): 65-76.
32. Widyaningsih N., W.H. Mohtar, N.I. Yussof,
M.D. Putri. 2022. „Impact
Of Large-Scale Social Restrictions On Transportation
Modes”. International Journal on Technical and
Physical Problems of Engineering 14(51): 216-221.
33. Khursheed S., F.A. Kidwai. 2023. “Level of Service based Performance Model of BLUE Line of Delhi Metro
post-COVID-19”. In: Proceedings of the 15th International Conference of Eastern Asia Society
for Transportation Studies (EASTS). Vol.14. 2023, September 4-7, Selangor, Malaysia.
Received 14.05.2024; accepted in revised
form 09.07.2024
Scientific
Journal of Silesian University of Technology. Series Transport is licensed
under a Creative Commons Attribution 4.0 International License