Page 100 - Kỷ yếu hội thảo khoa học quốc tế - Ứng dụng công nghệ mới trong công trình xanh , lần thứ 8

th
HỘI THẢO KHOA HỌC QUỐC TẾ ATiGB LẦN THỨ TÁM - The 8 ATiGB 2023 83

In the first section, students are asked to provide motorcycling, and bus. The independent variables
information about their travel behavior, that is the represent factors related to commuting behavior such
main mode of travel used as well as the travel time to as distance, time, and transportation cost, as well as
the university. individual student factors such as gender, academic
year, motorcycle ownership, and so on.
At the end of the first section are questions
regarding the reasons why students do not choose the The MNL model's features are independent of
current bus system and whether they are willing to irrelevant alternatives (IIA), which implies that the
shift to a new, efficient, and reliable bus system. ratio between two options depends only on the
characteristics of those two options and is unaffected
The second section of the questionnaire collected by the characteristics of other options. Additionally,
information on their school, gender, year of student, the MNL model offers a decision model that is
living status, motorcycle rider license, motorcycle focused on a person's characteristics rather than the
ownership, bicycle ownership, part-time job and alternatives.
student living cost. Students are also required to The multinomial logit (MNL) model is expressed as:
provide the name of their school and their residential
address. Based on this information, we determined the e V in
distance between the place of residence and the P (i) = V (1)
n
university campus using Google Maps. Then, distance  j C e jn

variables was transformed a categorical variable with n
4 scales named “DisTrans”. The travel time is directly with 0  P (i)  n 1 and  P (i) = 1
n

extracted from the survey results. The cost of j C n
transportation includes expenses for fuel and parking With Pn(i) is the probability that individual n
fee, as well as bus fares. The cost of using a choosing alternative i; i,j are alternatives in a choice
motorbike includes parking fees (2,000 VND) and set (in this study, they are walking, bicycle,
operating expenses (fuel costs), which are estimated motorcycle or bus); Vin is a systematic component of
based on the distance traveled and fuel costs per the utility function of individual n choosing
kilometer (500 VND/km). According to the typical alternative i;
fares for trips under 5 kilometers (6,000 VND) and The relationship attributes in Vin in this study is
over 5 kilometers (12,000 VND), the bus fare for assumed linear with parameters: Vin =k * xink.
users of public transportation is approximated. It has
th
been noted that the most popular form of k is parameters of Utility function with the k
transportation (motorcycles) has the highest financial attribute xink of individual n choosing alternative i.
expense. Then, ‘Cost’ variables was transformed a The Maximum likelihood method will be
categorical variable with 4 scales named “CostTrans”. conducted to assign a choice to each decision-maker
based on the maximization of deterministic utility.
B. Multinomial Logit model (MNL)
IV. RESULTS
A discrete choice modeling approach is employed
to develop a transport choice model, aiming to A. Descriptive statistics
comprehend people's travel behavior. In the theory of The total number of respondents received is 416,
random utility, every traveler selects an option that with 299 from UTE students and 318 from males
maximizes their personal utility. Different methods (Table 1). The majority (approximately 33%) of
are employed in the literature to identify the respondents are third-year students, followed by
relationship between exploratory variables and second-year students at 28.5%. First-year students
dependent variables. The MNL model is one of the accounted only 18% of respondents. Regarding to the
classical discrete choice models utilized in this paper. living status, approximately 68% students live with
The multinomial logit model is a method similar roommate in rented houses, whereas nearly 25% were
to binary logit regression, but it is applied when the living in their family-owned houses. Out of 416
dependent variable is a categorical variable with more students, 394 students own a two-wheeled vehicle
than two categories. In the context of this study, the license, however 262 those own a motorcycle and 74
multinomial logit model is used to predict the own a bicycle. From the sample, the number of
influence of factors related to commuting behavior as students with part-time jobs and those without is
well as the socio-economic factors of students on their approximately equal. The majority (approximately
choice of commuting mode in the city of Danang. The 73%) of students live with expenses ranging from 3
dependent variable in this case consists of four million to 5 million VND. A minority of students
commuting mode options: walking, cycling, spend less than 3 million VND (21.1%) and more

ISBN: 978-604-80-9122-4