Nghiên cứu dự báo nhu cầu vận tải hành khách cho các phương thức vận tải ở đô thị việt nam TT TIENG ANH

MINISTRY OF EDUCATION AND TRAINING UNIVERSITY OF TRANSPORT AND COMMUNICATIONS PHAN NGUYEN HOAI NAM RESEARCH FORECAST TRAVEL DEMAND ALLOCATION FOR THE MODE OF TRANSPORT IN URBAN AREAS IN VIETNAM Majors: TRANSPORT ORGANIZATION AND MANAGEMENT Code: 9.84.01.03 SUMMARY OF ECONOMIC DOCTORAL DISSERTATION HA NOI, 2021 Thesis is completed at: University of Transport and Communications Scientific guide group: PhD Nguyen Xuan Hoan Assoc Prof PhD Tran Thi Lan Huong Reviewer 1: Reviewer 2: Reviewer 3: The thesis will be defended in front of Examination Committee at University Level In: University of Transport and Communications At …… , ……, …… …… 2021 The thesis can be found in the Library Information Center of University of Transport and Communications INTRODUCTION THE URGENCY OF RESEARCH In recent years, the development of national economy has been coupled with high urbanization, which made the demand for development of transportation industry increased This leads to worsening of serious traffic jam and accidents, and environmental pollution in developed cities To solve the above problems, we need a sustainable urban planning However, making an urban transport plan is firstly based on traffic demand forecast in general and travel demand forecast in particular Identifying correctly the transport demand creates favorable conditions for building and developing infrastructure and preliminary determining of land funds for transportation in long terms, ensuring effective settlement of problems The accuracy of travel demand forecast result are main factors determining the success in urban transport development in the future Large cities in Vietnam, especially Hanoi and Ho Chi Minh City, are facing with traffic jams, accidents and serious environmental pollution Therefore, it is essential to prepare a transportation system development plan to meet the increasing demands Forecasting transport demand in general and travel demand in particular is the scientific basis for this transport development planning Forecasting the distribution of travel demand for modes of transport is essentially the third step in a four-step travel demand forecasting model that determines the proportion of people who choose a mode of transport On the other hand, the forecast of travel demand is often associated with urban transport planning projects and is only a small part of the project, so forecasting the distribution of travel demand for different modes of transport is not realistic being respected The forecasting models for allocating travel demand for transport modes in Vietnam's urban areas in recent times have inherited from existing models in the world and have been adjusted to suit the conditions in Vietnam However, with the rapid growth of economic and population which leads to the living conditions of residents have changed, the old model becomes unsuitable Therefore, it is necessary to have more complete studies on forecasting to overcome these problems On that basis, fellows chosen the theme "Research forecast travel demand allocation for the mode of transport in urban areas in Vietnam" to perform the doctoral thesis PURPOSE OF RESEARCH Analyzing factors affecting travel demand such as socio-economic characteristics, transport supply, urban transport infrastructure, behavior of traffic participants Determining the factors that directly affect the choice of means of transport in order to build a forecasting model for the distribution of travel demand for different modes of transport in urban areas in Vietnam based on inheritance and complementarity supplementing and adapting existing modern models in the world SUBJECTS AND SCOPE OF RESEARCH - Research objects: Studying predictive models of travel demand distribution and factors affecting travel demand - Scope of the study: The characteristics of the city affect the need in the study of travel demand For small cities, where the transport infrastructure still meets the travel needs of people, traffic jam and environmental pollution are not serious, researching the travel needs as well as re-planning the urban transport is not urgent yet However, for big cities, where large number of people participate in traffic, the transport infrastructure no longer meets the travel demand, traffic congestion and serious environmental pollution happening daily, forecasting travel demand and re-planning urban transport becomes more urgent than ever Besides, each city has different characteristics, which leads to different factors affecting travel demand, so it is not possible to cover all cities Ho Chi Minh City is the largest city in the country with a high population growth rate, leading to traffic congestion and serious environmental pollution, so it is really necessary to have studies to solve this situation Therefore, the thesis delves into the study of travel demand in Ho Chi Minh City, using data sources collected from actual surveys in Ho Chi Minh City to build and select a distribution forecasting model The travel demand for the modes of transport (specifically, the model predicting the probability of choosing a mode of transport of the person making the trip in Ho Chi Minh City for the basic trip from home) is suitable RESEARCH METHODS Based on research achievements of advanced countries, scientifically combining system theory, economic theory, econometric theory, transportation planning theory, the thesis uses a combination of qualitative and quantitative research methods as a basis for building and selecting a model to forecast the distribution of travel demand for each mode of transport RESEARCH CONTENTS The main content of the thesis research consists of chapters: - Chapter 1: Overview of the researched problems - Chapter 2: Theoretical basis for forecasting the distribution of travel demand for different modes of transport Transport modes in urban areas - Chapter 3: Approach method - Chapter 4: Research results on forecasting the distribution of travel demand for applied modes of transport in Ho Chi Minh City CHAPTER OVERVIEW OF THE RESEARCHED PROBLEMS In forecasting the distribution of travel demand for modes of transport in urban areas, there are many types of models used for analysis and forecasting According to the general or specific approach, they can be divided into two types: global (or general) and local model Global (or general) models includes of choice modelling of one-way or two-way transportation For this model, it is often used multi-factor linear regression model Local models are used for selection of each individual mode of transport and is based on the theory of mode of transport utility to the person making the trip Local models are commonly known as the Probit model and the Logit model and is classified as a discrete model In the overview, the author analyzes and evaluates the researches that have been done in the country and in the world, focusing on the form of mathematical models applied in those studies to see advantages and disadvantages of each type, solved problems in reality and existing gaps in the research 1.1 Researches in the world 1.1.1 “Pittsburgh Area Traffic Study” Program 1.1.2 Washington DC City Travel Demand Research Program 1.1.3 Study on Highway by US National Research Council 1.1.4 Study on mode choice of transport for two-way trips by Riga-Daugavpils transport company 1.1.5 Comparative study of models which classifying modes of transport 1.1.6 Study on urban transport planning by Michael D Meyer and Eric J Miller 1.1.7 Study on travel demand in Dhaka city, Bangladesh 1.1.8 Study on estimated value of travel time for a work trip 1.1.9 Research on traffic and expressway design 1.1.10 Research on vehicle selection decision in Nanjing, China 1.1.11 Study on theoretical framework for Logit model 1.1 12 Research on the choice of mode of transportation of workers in Chennai city 1.1.13 National transport research cooperation program countries, USA 1.1.14 Minal and Ch's study on transport mode division Ravi Sekhar 1.1.15 Research on transportation and regional growth of the University of Minnesota 1.2 Domestic research projects The theory of multi-factor linear regression model and its application in forecasting travel demand has long been mentioned in training program of the University of Transport However, in practice, the application of is model in travel demand forecast is often used in global model or in direct travel demand forecasting models For the 4-step forecasting model, the multi-factor linear regression model is mainly used to calculate the amount of derivative – attracting transport Then, this amount is allocated to types of transport modes All recent studies in Vietnam have not used multi-factor linear regression model in this step All forecasting models of urban traffic demand developed by foreign experts in large cities in Vietnam use the four-step forecasting method Predictive models are built based on existing models and adjusted for parameters by traffic survey data For forecasting the distribution of travel demand, most studies determine the travel demand by logit model of individual transport, then study the possibility of switching to public transport mode However, the variables used in the logit model and the calculation form of the conversion to public transport in the projects are different Beside the four-step forecasting model, some projects determine travel demand using the travel coefficient method 1.2.1 Project "Master plan for transportation development in Ho Chi Minh City until 2020" 1.2.2 Project "Master plan and feasibility study on urban transport in Ho Chi Minh City” 1.2.3 Feasibility study project of two priority metro lines City Ho Chi Minh City 1.2.4 Overall Urban Development Program in Hanoi Capital 1.2.5 Feasibility Study of Hanoi Urban Railway 1.2.6 Project of Metro Project – Line 1.3 Gaps in Research and propose research directions of the topic 1.3.1 General assessment Based on analysis of domestic and foreign studies related to forecasting the distribution of travel demand for different modes of transport in cities, we can see that most of the studies on forecasting travel demand in recent times use 4-step forecasting model This is a forecasting model that has proven to be more accurate and stable than previous direct travel demand forecasting models The principle of choosing in forecasting model is inherited and developed from the utility maximization principle in economic analysis, with the basic idea that the person making the trip will choose mode of transport in order to maximize the utility or benefit they can gain About the type of forecasting model applied exclusively to step (the step of allocating travel demand to modes of transport) in the 4-step forecasting model, which is also the research content of the topic, recent studies all use discrete models Discrete models have been applied in practice to forecast the distribution of travel demand (including linear probability models, Probit models and Logit models), due to the advantage of simplicity in computation Since there is no requirement on the condition of the random factor and the ability to forecast results with higher accuracy, the Logit model is considered to be the most suitable and commonly used in forecasting of allocated travel demand both in Vietnam and around the world nowadays Although the Logit model with the utility maximization principle is widely recognized and used in forecasting the distribution of travel demand in the world, each country and each city still have different characteristics, different number of transport modes and different affections on travel demand, so the Logit model often has to be adjusted to suit the studied area These adjustments are done more or less depends on the assessment of gaps in the study 1.3.2 Gaps in research Thus, learned from the studies of allocated travel demand forecast, author gives comments as below: a For overseas studies Overseas studies did in-depth research on the 4-step forecasting process in general and forecasting the distribution of travel demand in particular These studies have clarified the travel demand forecasting process according to the fourstep model, and through practical studies to gradually improve the theoretical system for predicting travel demand For forecasting the distribution of travel demand for modes of transport, each study is based on the analysis of the specific conditions of each territory, thereby provided or adjusted the forecast models suitably Theoretically, all studies have shown a lot of factors affecting the utility function, but in practical studies, most of the studies only focus on two factors, which are time and travel cost without any practical evidences for the influence of the remaining factors On the other hand, each country and each territory has its own characteristics, so the overseas research models are only valid for reference but cannot be applied rigidly in forecasting the distribution of travel demand for different modes of transport in urban areas in Vietnam b For domestic studies Urban traffic demand forecasting models developed by foreign experts forecast in large cities in Vietnam as well as travel demand forecasting models applied in Urban transport planning projects conducted by domestic experts all use the four-step forecasting method Predictive models are built based on existing models and adjusted for parameters by traffic survey data For forecasting the distribution of travel demand for modes of transport, due to the inheritance of available models that have been studied in the world, in the studies when building the utility function, the authors only study The two main factors are time and travel cost In fact, with specific conditions like in Vietnam, the choice of mode of transport may also depend on many other factors, such as the average income of the person making the trip or opportunity to use means of transport These factors need to be studied further On the other hand, in the current period, both Ho Chi Minh City and Hanoi are gradually forming elevated railway lines, however, there are not many studies on the model of distribution of travel demand for different modes of transport in the context of the emergence of new modes of transport, so this issue also needs to be studied more deeply and comprehensively 1.3.3 Research direction of the topic Based on the gaps of the research mentioned in Section 1.3.2, the research direction of the topic will focus on solving the following problems: - Analysis of influencing factors to the choice of mode of transport of the person making the trip to evaluate whether in addition to the time and cost factors, in the specific conditions in urban areas in Vietnam, what other factors can affect influence their choice of mode of transport - Based on analysis of influencing factors, building a logit model to forecast the distribution of travel demand for different modes of transport or a model predicting the probability of choosing a mode of transport of the person making the trip in conditions for the emergence of a new mode of transport is the elevated railway for Ho Chi Minh City The research question is: - Does the difference among modes of transport affect the choice decision of the person making the trip? - How does the average monthly income of the person making the trip affect the choice of mode of transport? - Does the ratio of trip expenses / average monthly income of trip takers affect their decision to choose a mode of transport? - In fact, there are modes of transport where the time outside the vehicle is very large, so is this an obstacle for the person making the trip when making the decision to choose the mode of transport? - How does the availability of personal vehicles in each family making the trip influence their choice decision? - In the event of the emergence of a new mode of transport, how will the pattern of distribution of cross-border demand for different modes of transport change? Conclusion of Chapter Forecasting travel demand in general and forecasting the distribution of travel demand for modes of transport in urban areas in particular (forecasting the choice of mode of transport of the person making the trip) is the first step important in the process of urban transport planning Researches on forecasting in this field are many in many countries around the world, but each country has its own characteristics in terms of culture, socio-economics, transportation system so the models Forecast models applied in different countries are difficult to apply to each other Therefore, when researching predictive models, it is necessary to make appropriate adjustments for each specific condition and situation In order to build a forecasting model for the distribution of travel demand for transport modes in urban areas in Vietnam, it is necessary to study the models that have been applied in the world and in Vietnam itself From there, we can make adjustments, improve the model or build a suitable new model Chapter of the thesis focuses on studying a number of works that have been done in the world and in Vietnam related to forecasting the distribution of travel demand for transport modes in urban areas On that basis, the thesis points out the gaps in the research that can still be exploited in the process of building a forecasting model to allocate travel demand for modes of transport in urban areas in Vietnam and propose research direction of the thesis CHAPTER THEORETICAL BASIS FOR FORECASTING THE DISTRIBUTION OF TRAVEL DEMAND FOR DIFFERENT MODES OF TRANSPORT TRANSPORT MODES IN URBAN AREAS Based on analysis and research orientation in chapter 1, before going into modeling studies forecast to allocate travel demand for modes of transport in urban areas, chapter will clarify the theoretical bases for the research process such as the theory of the behavior of the person making the trip and the theory of forecasting travel needs 2.1 Behavior of the person making the trip The need for travel is a secondary need that originates from other needs in human life, such as to go shopping, people have the need to move from home to home supermarkets or shopping malls Transport services are considered as a special type of goods, so the behavior of using transport services is also considered as the behavior of consuming tangible goods and the person making the trip can be called a person transportation consumption There are many different views on consumer behavior, but when analyzing consumer behavior, scientists usually analyze it from three points of view, namely: a psychological point of view, a marketing point of view and an economic point of view 2.1.1 The behavior of the person making the trip from the psychological point of view The behavior of choosing a means of transport is a unified whole consisting of internal and external factors that have close relationships with each other Internal factors can be understood as personal characteristics, psychological characteristics, preferences of consumers, external factors are family characteristics, living environment, activities of that person 2.1.2 Traveler’s behavior from the marketing point of view From a marketing point of view, the process of deciding on the means of transport of the trip maker is a series of stages that the trip maker goes through in deciding your options 2.1.3 Traveler’s behavior from an economic point of view According to an economic point of view, people making the trip will decide their behavior in order to maximize their utility or maximize their utility 2.2 Theory of travel demand forecasting 2.2.1 Urban travel and passenger transport demand Travel demand is understood as the largest possible number of trips by a person or a group of people or in a region, an area in a certain period of time To fulfill their desire to travel, human will choose a mode suitable for specific circumstances, which can be on foot, by private means of transport or by public transport To serve the modeling work later, the thesis categorizes the factors affecting the choice of transport mode into groups: The person making the trip and the characteristics of the person making the trip; availability of alternative modes of transport; economic - technical characteristics of modes of transport and decisionmaking principles of the person making the trip 2.2.2 Analysis and forecasting of urban mobility demand distribution Urban commuting demand models are often based on hypotheses related to the behavior of urban commuters, which can be obtained through survey of people's habits The basic method for analyzing urban travel demand is the theory of demand in microeconomics In particular, the principle of utility maximization is often used to analyze the trip choice of the trip maker There are two approaches to modeling urban travel demand, namely: the direct approach and the sequential (indirect) structural choice model approach According to the direct approach, the travel demand function model is built according to the influencing factors and the number of trips is estimated or predicted directly from this model Meanwhile, the indirect modeling approach assumes that there are many possible choices of the trip operator, which leads to the structure of a series of choice models and their combination to predict the total amount of trips with different purposes Conclusion of Chapter The process of choosing a mode of transport of the person making the trip depends on many factors such as the characteristics of the person making the trip, the characteristics of the transportation system and the mode of transport Besides, Random effects can also radically change their choice In order to be able to build a model for forecasting the choice of transport mode or a model of allocating travel demand for the modes of transport of the trip-performers, before proposing the model, it is necessary to have Studies on the theoretical basis of travel demand forecasting Chapter of the thesis analyzes the behavior of the person making the trip from the psychological, marketing and economic points of view, and at the same time clarifies the theoretical basis of the process of forecasting travel demand in general and forecasting choice choose a mode of transport for the person making the trip in particular In addition, chapter also offers two approaches to modeling travel demand and clarifies the process of forecasting travel demand according to the four-model step Based on the analysis of the behavior of choosing a mode of transport of the trip takers described in chapter 2, it can be seen that whether approaching from a psychological point of view, from a marketing point of view or from an economic point of view, then the decision to choose the mode of transport of the person making the trip has similarities It is the decision to choose the mode of transport of the person making the trip which is influenced by a set of influencing factors and the person making the trip will choose the mode of transport in the way that they think is reasonable corresponding to that set of influencing factors Of the two approaches to modeling travel demand, as assessed in section 1.3.1, the sequential structural approach (typically the 4-step model for forecasting travel demand) has the advantage advantages over the direct approach Besides, in the approach to the behavior of choosing a mode of transport of the trip takers from an economic point of view, this choice decision is made through the evaluation of the utility function with the principle of maximizing satisfaction of the person making the trip The construction of the utility function is also simple and suitable for later modeling rather than the emotional judgments that may appear when analyzing the behavior of choosing a mode of transport of people who take the trip according to their preferences psychology and marketing perspective Therefore, the method of sequential structural choice approach in modeling travel demand is based on an economic point of view in analyzing the mode of transport choice behavior of the selected trip takers used in the next chapters of the thesis 11 collected based on the questionnaire according to the new 5-level Liker scale shows the importance of influencing factors without specifying the specific influence of each factor in actual conditions Therefore, it is necessary to build a forecasting model for the distribution of travel demand based on large sample data, to be examined practically and in more detail, in order to affirm in practice, the importance of employees influencing factors as well as specifying the level of influence of factors in specific actual conditions The survey data set on travel demand in Ho Chi Minh City surveyed by SUD company satisfies the above requirements, so it is selected to be used in the next research step 3.2 Data for forecasting the distribution of travel demand Survey data conducted by SUD company based on household interviews by stratified random sampling method conducted at the beginning of 2019 in the city Ho Chi Minh City with 12432 people over years old participating in the survey The structure of trips out of a total of 43130 surveyed trips can be described as below: a Allocating travel demand by mode of transport Table 3.19 Structure of trips by mode of transport Mode Number of trips (trips) Rate (%) Walk Motorcycle Bike Car Bus Taxi MotorcyOthers cle Taxi 4610 31 214 2075 866 2251 1503 138 473 43130 10.69 72.37 4.81 2.01 5.22 3.48 0.32 1.10 100.00 Total (Source: Calculation from SUD survey data) b Allocating travel demand by mode of transport - travel purpose Table 20 Trip structure by mode - purpose (unit: trip) METHODS OF TRANSPORTS PURPOSE WALK GO to WORK Go to shool Return home Go to eat Business trip Personal job SOCIAL SHOPPING OTHERS TOTAL 630 117 1842 652 127 167 477 589 4610 MOTOR- CYCLIN MOTOR XEBUYT CYCLE G 7614 748 13811 2583 159 559 823 2585 2332 31214 230 394 960 103 28 45 199 112 2075 222 365 95 18 18 23 53 70 866 523 315 1039 23 57 75 128 87 2251 TAXI 414 627 164 24 19 206 40 1503 motorbik OTHER e 13 64 10 20 17 138 102 70 166 14 49 4 19 45 473 TOTAL 9748 1655 18874 3644 250 823 1157 3687 3292 43130 (Source: Calculation from SUD survey data) 3.3 Selection of forecasting model 3.3.1 Evaluation and choosing the type of model to forecast the distribution of travel demand for modes of transport in urban areas Today, most of the studies follow the discrete approach because this direction has many advantages compared to the traditional method holistic approach in modeling the decision-making behaviors of a group of trip-makers 12 Based on these assessments and the analysis of the advantages and disadvantages of the predictive models for choosing the mode of transport of the past and present trip takers in Chapter 1, the polynomial logit model can be considered as is a suitable model for predicting the choice of mode of transport of trip takers 3.3.2 The polynomial logit model in forecasting the distribution of travel demand for modes of transport (forecasting the choice of transport mode) Polynomial (or multi-selection) logit regression is an extension logit regression method (where the dependent variable is a binary variable) into classification problems Where, the dependent variable is categorical and there are more than two possible discrete outcomes (more than two categories), the independent variables have real, binary or categorical values Probability of a certain outcome: ( ) ∑ Where: is the utility function of the outcome j Utility function structure In terms of form, we can express: (3 26) (3.27) Where: Uit: is the real utility of the option i for the person taking the trip t ) : is the observable utility (systematic or deterministic utility), ,( which is observable and estimated by the analyst can be expressed as a function of Xi and St or: ( ) (3 28) : is the error or utility unknown to the analyst Therefore, the utility model can be expressed as: ( ) (3 29) Evaluation and selection of influencing factors in the utility function With the model estimation results in section 3.1.5 we can arrange the influencing factors in descending order of the absolute value of the corresponding parameters, to evaluate the importance of the influencing factors With the results in the table below, the influencing factors can be divided into groups according to the level of influence - Group 1: factors that have a lot of influence: - Group 2: factors that have little influence with coefficients ranging from 0.04 to 0.07 Table 3.32 Order of influence of factors No Variable TG CP CT TN CH CL Coefficient 0.191 0.184 0.130 0.067 0.067 0.064 TT 10 11 12 Variable NN T CS KC SHBL GTI Coefficient 0.059 0.057 0.056 0.054 0.050 0.042 13 Based on the results obtained on these influencing factors, the influencing factors by the selected author to be used in the forecasting model include: - Three factors in group 1: trip time trips, trip costs and expense-to-income ratios - The top two factors in group are income and the opportunity to use personal means of transport Conclusion of Chapter A model for forecasting travel demand in general and predicting the choice of transport mode of trip-makers in particular depends on many factors, two of which are important: The important determinant of success or failure when building a model is a research method to build a model and a survey data set for parameter estimation for the model Chapter solves the two above problems by clarifying the research method used in the process of building a predictive model to choose the mode of transport of the trip operator and how to collect and process data in that process The study of influencing factors is designed into phases: preliminary research, experimental research and formal research with the process and implementation time described clearly in chapter The first two phases aim to complete the study improve the survey questionnaire as well as identify the factors that can influence the decision of choosing a mode of transport of the trip takers The final stage performs the task of assessing and determining the factors that officially affect the decision of the person making the trip to choose a mode of transport The results of the exploratory factor analysis and the estimation results of multiple linear regression models show that the factors affecting the decision to choose the mode of transport of the person making the trip to Ho Chi Minh City are as originally intended for the study The observed variables are separated into 12 separate factors and each factor has a different influence on the decision to choose the mode of transport of the person making the trip to Ho Chi Minh City Besides, chapter also clarifies the method of data collection and processing for the research process The questionnaire was built according to the purpose of studying the influencing factors that were oriented in the thesis Based on the number of questions in the questionnaire and the basis of statistical theory, the author provides sample size and sample structure for the data survey process The method of data survey is household interview and workplace interview The collected data will be processed by popular statistical software such as SPSS or STATA Based on assessment of influencing factors, the author chooses a forecasting model which is a polynomial logit model and continues to build a theoretical framework for a polynomial logit model to determine the probability of distribution of travel demand 14 CHAPTER RESEARCH RESULTS ON FORECASTING THE DISTRIBUTION OF TRAVEL DEMAND FOR APPLIED MODES OF TRANSPORT IN HO CHI MINH CITY In Chapter 3, the author reviews, selects form of forecasting models and The influenced factor that can be used in the model as well as build a theoretical framework for the model to forecast the distribution of travel demand for modes of transport in urban areas However, this model needs to be verified to ensure applicability in practice Therefore, in Chapter 4, the author uses a large sample data set surveyed by SUD Company in Ho Chi Minh City as a representative to test practically the polynomial Logit model selected in Chapter 3, as well as the assumptions made in the study 4.1 Introduction to the study area 4.1.1 Socio-economic development situation in Ho Chi Minh City With over 8.2 million residents living in the area and contributing over 20% of the national GDP, the city Ho Chi Minh City is considered the largest and most developed city in the country, surpassing the capital Hanoi According to the population projection results, the population of Ho Chi Minh City will continue to increase and is estimated to reach about 9.2 million people in 2020 to 10 million people in 2025 4.1.2 Transportation issues a The supply capacity of the transport infrastructure system is still limited b Increased ownership and use of motor vehicles c Severe traffic jam d Low efficiency of public air transport 4.2 Building a model to predict the probability of choosing a mode of transport in Ho Chi Minh City A model predicting the probability of choosing a mode of transport of people making a trip in Ho Chi Minh City was built based on inheriting previously studied models and then adjusted to suit the current situation of travel demand in Ho Chi Minh City, considered the emergence of new modes of transport which is elevated railway (METRO) The thesis exploits a large sample survey data surveyed by SUD company and the model is limited in the number of influencing factors based on the data on the influencing factors that have been actually surveyed In addition, due to the combination of actual survey data and hypothetical survey data (the hypothetical survey data is only conducted with the basic trip from home) with the purpose of studying the case of the emergence of different modes of transport Since the new load is the elevated railway, the probability model of the transport mode choice of the person making the trip to Ho Chi Minh City continues to be limited to only basic trips from home 4.2.1 Predictive model type The proposed model is a polynomial logit model of the following form: 15 ( ) ∑ (4 1) In which: ): Probability that the person making trip i chooses method j + ( + :function of individual i for method j + k: number of modes of transport + e: Natural base + is the observable utility function - : random factor 4.2.2 Approach to predictive modeling As suggested in Section 3.2.2, the variables expected to be used in the study include: - CH: describe the opportunity to use personal means - TG: time trip, trip time can be split into time on vehicle (TGT) and time off vehicle (TGN) - TNT: ratio of out-of-vehicle time to total trip time - CP: cost for trip - TN: average monthly income - LTN: natural logarithm of income (this is a variation of TN to improve the model if possible) - CT: ratio of trip expenses to average monthly income 4.2.3 Building a predictive model The process of building and improving the model starts with a common model commonly used in research on choice of transport modes with two familiar independent variables: trip time and cost Then the adjusted variables are added to the model to obtain the model that best fits the data set used, including the statistical fit and is compatible with the theory of choosing behavior of mode of transport by the person making the trip From these adjustments, model MH5 was selected to be used in the study The utility function in the model has the form: (4.2) Model estimation results MH5 shown in Table 4.17 The MH5 model is extended in the case of research on the choice of transport mode when elevated railway (METRO) appeared, the survey data used to estimate the MH55 model is combined with the Hypothetical survey data collected in the same survey gives the model estimation results in the case of METRO (MH METRO model) as shown in Table 4.19 16 Table 4.17 Estimation results model MH5 Alternative-specific conditional logit Case variable: IDtrip Number of obs Number of cases = = 99456 12432 Alternative variable: METHODS Alts per case: = avg = max = 8.0 Log likelihood = -8211.3603 Wald chi2(18) Prob > chi2 = = 3505.28 0.0000 -LC | Coef Std Err z P>|z| [95% Conf Interval] -+ -MODES | TGT | -.181658 0051602 -35.20 0.000 -.1917719 -.1715442 TGN | -1.451482 057328 -25.32 0.000 -1.563843 -1.339122 CP | -.0063868 0014072 -4.54 0.000 -.0091449 -.0036287 CT | -11.24384 1.808351 -6.22 0.000 -14.78815 -7.699542 -+ -WALK | CH | -2.345925 2077353 -11.29 0.000 -2.753079 -1.938772 TN | -.0000186 00002 -0.93 0.351 -.0000579 0000206 _cons | 3.960927 1870932 21.17 0.000 3.594231 4.327622 -+ -OTHERS | CH | 1419767 3492539 0.41 0.684 -.5425483 8265016 TN | -.0001332 0000341 -3.91 0.000 -.0002 -.0000663 _cons | -2.85582 3178801 -8.98 0.000 -3.478854 -2.232787 -+ -CAR | CH | 3.511957 2668024 13.16 0.000 2.989034 4.03488 TN | 000371 000018 20.63 0.000 0003357 0004062 _cons | -9.067827 3458089 -26.22 0.000 -9.7456 -8.390054 -+ -TAXI | CH | -2.417663 1765795 -13.69 0.000 -2.763752 -2.071574 TN | 9.68e-06 0000208 0.46 0.642 -.0000311 0000505 _cons | 6284157 1792199 3.51 0.000 2771511 9796802 -+ -XEBUS | CH | -1.961617 1933294 -10.15 0.000 -2.340536 -1.582698 TN | -.0001344 0000201 -6.70 0.000 -.0001737 -.000095 _cons | 43.5422 1.743117 24.98 0.000 40.12575 46.95865 -+ -BICYCLE | CH | 1.040565 1795437 5.80 0.000 6886662 1.392465 TN | -.0004381 0000213 -20.61 0.000 -.0004798 -.0003965 _cons | -.5753208 1650979 -3.48 0.000 -.8989067 -.2517349 -+ -MOTORBIKE | (base alternative) -+ -MOTORCYCLE TAXI | CH | -4.781718 7377538 -6.48 0.000 -6.227689 -3.335747 TN | -.0003077 0001048 -2.93 0.003 -.0005131 -.0001022 _cons | -.6394166 5492362 -1.16 0.244 -1.7159 4370667 17 Table 4.19 Estimation results of MHMETRO model Alternative-specific conditional logit Case variable: IDtrip Alternative variable: METHODS Number of obs = 111888 Number of cases = 12432 Alts per case: = avg = 9.0 max = Wald chi2(20) = 3872.82 Log likelihood = -12978.62 Prob > chi2 = 0.0000 -LC | Coef Std Err z P>|z| [95% Conf Interval] -+ -MODES | TGT | -.1168913 0029462 -39.68 0.000 -.1226657 -.111117 TGN | -1.276916 0650528 -19.63 0.000 -1.404417 -1.149415 CP | -.0396926 0022326 -17.78 0.000 -.0440684 -.0353167 CT | -6.05722 1.267129 -4.78 0.000 -8.540748 -3.573692 -+ -WALK | CH | -2.516121 1986517 -12.67 0.000 -2.905471 -2.126771 TN | -.0000265 000019 -1.39 0.164 -.0000637 0000108 _cons | 2.966935 1671368 17.75 0.000 2.639353 3.294517 -+ -OTHERS | CH | 0920553 3824375 0.24 0.810 -.6575084 8416189 TN | -.0002004 0000387 -5.17 0.000 -.0002763 -.0001245 _cons | -2.894652 3486402 -8.30 0.000 -3.577974 -2.211329 -+ -METRO | CH | -.499287 0962334 -5.19 0.000 -.687901 -.3106729 TN | 3.51e-06 8.72e-06 0.40 0.687 -.0000136 0000206 _cons | 11.75996 6581428 17.87 0.000 10.47003 13.0499 -+ -CAR | CH | 3.017437 2623291 11.50 0.000 2.503282 3.531593 TN | 0002155 0000142 15.19 0.000 0001877 0002433 _cons | -7.533833 3123855 -24.12 0.000 -8.146098 -6.921569 -+ -TAXI | CH | -2.242785 1951894 -11.49 0.000 -2.62535 -1.860221 TN | 0000587 000018 3.27 0.001 0000235 0000939 _cons | 1.293522 1791986 7.22 0.000 9422993 1.644745 -+ -BUS | CH | -2.116661 2180381 -9.71 0.000 -2.544008 -1.689314 TN | -.0002002 0000238 -8.40 0.000 -.000247 -.0001535 _cons | 38.34651 1.971912 19.45 0.000 34.48164 42.21139 -+ -BICYCLE | CH | 7784058 1852237 4.20 0.000 4153741 1.141437 TN | -.0004706 0000224 -21.02 0.000 -.0005144 -.0004267 _cons | -.7091006 1701822 -4.17 0.000 -1.042652 -.3755496 -+ -MOTORBIKE | (base alternative) -+ -MOTORCYCLE TAXI | CH | -4.736579 7929821 -5.97 0.000 -6.290795 -3.182363 TN | -.000229 0001044 -2.19 0.028 -.0004336 -.0000244 _cons | 0113391 5564947 0.02 0.984 -1.079371 1.102049 18 4.3 Discussion about the results 4.3.1 Importance of influenced factors Trip time is a factor has the most important influence on the decision to choose the mode of travel of the person making the trip, followed by the cost of the trip and the ratio of expenses to income 4.3.2 Influence of factors a Effect of time on vehicles Table 4.21 Results of analysis of marginal variation of time on vehicles probability Methods Standard error Statistics Z P>Z change Motorbike -0.025364 0.000622 -40.76 0.000 Metro -0.021995 0.000572 -38.44 0.000 Bus -0.002694 0.000195 -13.79 0.000 Bicycle -0.002264 0.000165 -13.75 0.000 Others -0.001065 0.000115 -9.24 0.000 Taxi -0.000813 0.000096 -8.43 0.000 Car -0.000696 0.000082 -8.48 0.000 Walk -0.000124 0.000014 -8.60 0.000 Motorcycle taxi -0.000040 The results indicate the time on the vehicle increases, the probability of selection of modes of transport reduced The data also show that when the time on the vehicle changes, it affects the choice of modes of transport, of which the strongest impact is on the choice of motorbike, then the choice of a new mode which is elevated railway b Effect of time outside the vehicle The results are all negative, indicating that when the time outside the vehicle increases, the probability of choosing a mode of transport decreases The data also show that when the time outside the vehicle changes, it affects the choice of modes of transport, in which the strongest impact is on the choice of motorbikes, then the choice of buses, bicycles and bicycles, and new mode of transport which is the elevated railway Table 4.22 Results of analysis of marginal variation in time outside the vehicle probability Methods Standard error Statistics Z P>Z change Motorbike -0.277073 0.014213 -19.49 0.000 Bus -0.029424 0.002305 -12.77 0.000 Bicycle -0.024733 0.002181 -11.34 0.000 Metro -0.021995 0.000572 -38.44 0.000 Others -0.011638 0.001366 -8.52 0.000 Taxi -0.008878 0.001133 -7.83 0.000 Car -0.007606 0.000965 -7.88 0.000 Walk -0.001356 0.000201 -6.73 0.000 Motorcycle taxi -0.000442 - 19 c Effect of trip cost The data shows that the effect of trip cost has the greatest impact on the choice of motorbike, followed by the elevated railway All other methods are affected by cost but the effect is small The results of the probability fluctuations are negative, consistent with the theory that when the trip cost increases, the probability of choosing a mode of transport will decrease Table 4.23 Results of Analysis of Marginal Variation in Trip Costs Change in Standard Method Z P>Z Probability Statistical Motorbike -0.0086130 0.0004800 -17.95 0.000 Metro -0.0074690 0.0004270 -17.50 0.000 Bus -0.0009150 0.0000830 -11:05 0000 Bicycle -0.0007690 0.0000710 -10.77 0.000 Others -0.0003620 0.0000440 -8.28 0.000 Taxi -0.0002760 0.0000220 -12.69 0.000 Car -0.0002360 0.0000300 -7.83 0.000 Walk -0.0000420 0.0000065 -6.44 0.000 Motorcycle Taxi -0.0000140 d Effect of cost-to-income Cost-to-income ratio has the greatest influence on the probability of choosing a motorcycle, followed by elevated railway The third most affected is buses The results of the change in probability are all negative, showing that as the cost-to-income ratio increases, the probability of choosing a mode of transport decreases Table 4:24 Results of the analysis of fluctuations in the marginal rate on income expenses Change Method Standard error Statistic Z P> Z probability Motorbike -1.314330 0.273995 -4.80 0.000 Metro -1.139780 0.237262 -4.80 0.000 Bus - 0.139577 0.030897 -4.52 0000 Bicycle -0.117324 0.026484 -4.43 0000 Others -0.055208 0.013043 -4.23 0.000 Taxi -0.042112 0.009301 -4.53 0.000 Car -0.036079 0.008609 -4.19 0.000 Walk -0.006432 0.001587 -4.05 0.000 Motorcycle Taxi -0.002098 - - - e The influence of income The influence of income on the probability of choosing a mode of transport according to the calculation results shows that: groups that are strongly affected include: motorbikes, bicycles, buses, elevated railways; Less affected groups include: cars, taxis, motorbike taxis and other modes This result does not show a clear impact of income on the form of walking 20 Table 4:25 Results of analysis of marginal fluctuations Income Change Method Standard error statistic Z P> Z probability Motorbike 0.000009100 0.00000170 5:50 0.000 Bicycle -0.000009000 0.00000043 -21.15 0.000 Bus -0.000004400 0.00000049 -8.95 0.000 Metro 0.000004200 0.00000160 2.61 0.009 Others -0.000001700 0.00000030 -5.66 0.000 Car 0.000001400 0.00000015 9.03 0.000 Taxi 0.000000500 0.00000013 3.83 0.000 Motorcycle taxi -0.000000075 Walk -0.000000014 0.00000002 -0.69 0.488 f Effect of opportunity to use personal vehicle Table 4.26 Results of analysis of marginal variation in opportunity to use personal vehicle Method Change in Standard error Statistical Z P>Z probability Motorbike 0.109858 0.018461 5.95 0000 Metro -0.084987 0.017887 -4.75 0.000 Bus -0.046150 0.005029 -9.18 0.000 Car 0.019047 0.001952 9.76 0.000 Bicycle 0.018565 0.003600 5:16 0.000 Taxi -0.014574 0.001993 -7.31 0.000 Walk -0.002503 0.000378 -6.62 0.000 Motorcycle Taxi -0.001585 Others 0.002329 0.003464 0.67 0.501 The effect of the opportunity to use personal vehicles according to the calculation results is in line with the theory, the mark of the results shows that when the opportunity to use private vehicles increases, the probability of choosing bicycles, motorcycles and cars increases, and the probability of public and semipublic transport options decreases However, the results not clearly show the effect of the opportunity to use personal means of transport on the probability of choosing “others” mode of transport 4.3.3 Impact of the opportunity to use motorcycles on the choice of public transport Currently, the number of motorcycle trips is accounting for a large proportion of the total daily trips of people and motorbikes are considered one of the main causes of traffic jams In order to assess the individual impact of the opportunity to use motorcycles on the probability of selecting cm variable modes of transport (describing the opportunity to use motorcycles) separated from the CH variable and XEMAY is selected as the base property or comparative criteria in the model, the results are obtained as table 4.27 All parameters associated with CM variables for modes of transport bear negative marks indicating that when the opportunity to use motorcycles increases, 21 the probability of choosing a motorcycle will increase relative to the probability of choosing the remaining modes of transport, and obviously that when the opportunity to use motorcycles decreases, this probability will decrease Thus, according to the results obtained from the model, if there are solutions to limit the use of motorcycles reasonably, the percentage of trips by motorbike will decrease and the rate of trips by other means of transport will increase Table 4.27 Results of MHMETRO-BAESE-MOTORBIKE model Alternative-specific conditional logit Case variable: IDtrip Alternative variable: METHODS Number of obs Number of cases = 111888 = 12432 avg = 9.0 max = Wald chi2(20) = 3946.07 Log likelihood = -12906.295 Prob > chi2 = 0.0000 -LC | Coef Std Err z P>|z| [95% Conf Interval] -+ -Methods | TGT | -.1166965 002966 -39.34 0.000 -.1225097 -.1108832 TGN | -1.262047 062818 -20.09 0.000 -1.385168 -1.138926 CP | -.0396217 0022495 -17.61 0.000 -.0440306 -.0352128 CT | -7.426039 1.31045 -5.67 0.000 -9.994474 -4.857603 -+ -WALK | TN | -.0000229 0000183 -1.25 0.211 -.0000587 000013 CM | -2.476635 2036476 -12.16 0.000 -2.875777 -2.077494 _cons | 2.73571 1536292 17.81 0.000 2.434602 3.036818 -+ -OTHERS | TN | -.0001247 0000374 -3.34 0.001 -.000198 -.0000514 CM | -2.73013 3702996 -7.37 0.000 -3.455904 -2.004356 _cons | -1.140959 2584068 -4.42 0.000 -1.647427 -.6344911 -+ -METRO | TN | 0000111 8.74e-06 1.27 0.206 -6.07e-06 0000282 CM | -1.226004 1020559 -12.01 0.000 -1.42603 -1.025979 _cons | 12.12015 6357549 19.06 0.000 10.87409 13.36621 -+ -CAR | TN | 0002771 0000145 19.11 0.000 0002487 0003055 CM | -.6271784 3658749 -1.71 0.086 -1.34428 0899233 _cons | -4.317707 3290125 -13.12 0.000 -4.96256 -3.672854 -+ -TAXI | TN | 0000469 0000179 2.63 0.009 0000119 0000819 CM | -2.21092 2022136 -10.93 0.000 -2.607252 -1.814589 _cons | 1.187137 1721961 6.89 0.000 8496385 1.524635 -+ -BUS | TN | -.0001822 0000237 -7.69 0.000 -.0002287 -.0001358 CM | -2.407256 2328483 -10.34 0.000 -2.86363 -1.950882 _cons | 37.87775 1.897837 19.96 0.000 34.15806 41.59745 -+ -BICYCLE | TN | -.0003332 0000223 -14.95 0.000 -.0003769 -.0002895 CM | -3.869335 199398 -19.41 0.000 -4.260148 -3.478522 _cons | 2.113294 1289034 16.39 0.000 1.860649 2.36594 -+ -MOTORBIKE | (base alternative) -+ -MOTORCYCLE TAXI | TN | -.0001984 0001035 -1.92 0.055 -.0004012 4.39e-06 CM | -6.289812 8755357 -7.18 0.000 -8.005831 -4.573794 _cons | 298146 5031399 0.59 0.553 -.6879901 1.284282 22 4.4 Some petitions - Use the MHMETRO model to forecast the probability of selecting the mode of transport of the person making the trip in Ho Chi Minh City - Because the cost-to-income ratio affects the probability of choosing the mode of transport of the person making the trip in Ho Chi Minh City, to improve the use of VTHKCC in the city, it is necessary to take measures to reduce the fare of VTHKCC - To reduce the rate of motorbike choice compared to other modes of transport, solutions are needed to limit people's opportunities to use motorcycles - According to the results of the study, the time outside the vehicle also affects the decision to choose the mode of transport of the person making the trip, the greater the time, the more hinders the decision to choose the mode of transport Therefore, measures should be taken to increase the density of the public transport network in order to reduce the time spent outside the vehicle, thereby attracting an additional number of people making the trip to use public transport Conclusion of Chapter Chapter of the thesis analyzed the travel needs of the person making the trip in Ho Chi Minh City, thereby showing the main characteristics of the travel needs of the people of the city Based on the analysis of the results obtained from the linear regression model in chapter 3, the author evaluates and selects the main factors influencing the decision to choose the mode of transport of the person making the trip and uses these factors to build a forecast model for the allocation of travel needs for modes of transport The influence of factors continues to be studied in more detail in the pollinate logit model to determine the probability of choosing the mode of transport of the person making the trip in Ho Chi Minh City The calculation of the polymerize Logit model on the set of data provided by SUD company aims to verify once again the research hypotheses on the influencing factor and determine the model of forecasting the probability of choosing the mode of transport for the person making the trip in Ho Chi Minh City in case of taking into account the appearance of the method New mode of transport After the process of researching improvements and adjustments, the results of chapter give a model of predicting the probability of choosing a mode of transport for the person making the trip in the city The study of the expansion of the model in the event of a new mode of transportation, Metro, was also conducted simultaneously The estimated results of the forecast model of transport selection for people making a trip to Ho Chi Minh City in the conditions of the emergence of a new mode of transport suggest that this model can be used for future forecasting Through the study results, chapter of the thesis also proposes some recommendations when using this forecast model 23 CONCLUSION The study forecasting the allocation of travel needs to modes of transport in urban Vietnam is an inseparable task from transport planning The existence of urban traffic in Vietnam such as traffic jams, pollution, noise causes many bad consequences that people suffer and require to be solved The main cause of the problem is the mechanical growth rate of the urban population is too fast, exceeding the responsiveness of the transport infrastructure Another reason is that previous models of travel demand forecasting may not clearly and accurately reflect the fluctuations in travel demand in general and travel demand for each mode of transport in particular Faced with this context, the thesis studies influencing factors and builds a forecast model for allocating travel needs to urban modes of transport in accordance with real conditions in Vietnam The thesis has a number of new contributions that are as follows: - For the four-step travel demand forecasting model used in the thesis, the author of the system of forecasting methods allocates travel needs for each mode of transport, assesses the advantages and disadvantages of each method to be the basis for selecting the forecast model In addition, the thesis also assesses the factors affecting the decision of traffic participants in the world in general and in Vietnam in particular and proposes a forecast model suitable to conditions in Vietnam - Based on the assessments and data collected, the thesis clearly indicates the reasons for choosing a model to forecast the allocation of travel needs for each mode of transport in major cities in Vietnam - The model of forecasting the probability of selecting the mode of transport of the person making the trip (specifically applied to Ho Chi Minh City) has taken into account the emergence of the new mode of transport is the elevated railway (Metro) In addition to the results achieved, there are some existences that need further study: - The quality of public transport services that are first and foremost expressed in the comforts of public transport means has influenced the decision to choose the mode of transport, but the issue of the quality of transport services has not been considered in the forecast model of allocating travel needs to urban modes of transport - The forecast model has not studied in detail the effect of the influence of government policies in general and the policy of limiting personal vehicles in particular on the probability of people choosing modes of transport - The model is built on the assumption that the Metro network covers all survey lines However, in fact, no line has been put into operation and is gradually completing each line, so further research is needed on the fluctuations in travel demand on each metro line that is about to operate - The development direction in the forecast model allocating travel needs to transport methods in the world in the current period is to use the cage logit model with the condition that it is possible to group the same types of transport methods However, currently in Vietnam, the Metro system is not complete and there is not 24 enough database to make forecasts on the cage Logit model, so the cage logit model needs to be studied more deeply to serve the forecasting of future travel demand Due to the limited time and knowledge, the thesis is not immune to shortcomings and is eager for sincere comments from scientists, teachers and other individuals who are interested in the thesis Finally, the author would like to sincerely thank the teachers and teachers for their dedication in guiding the author to complete his thesis The author also thanked the teachers in the Faculty of Transport - Economics, University of Transport and friends who enthusiastically supported in the process of completing the thesis LIST OF THE PUBLICATIONS Phan Nguyen Hoai Nam, Do Minh Ngoc, Pham Ngoc Hai (2017), “Analyzing the inadequacies in the model of forecasting demand for passenger transportation arising in Vietnam”, Transport and Communication Journal, December 2017 issue, pp 174-177 Phan Nguyen Hoai Nam, Do Minh Ngoc, Pham Ngoc Hai (2017), “Choosing the model of forecasting demand arising - attraction in passenger transportation for Ho Chi Minh city”, Scientific research project at University of Transport and Communications, code No: T2017 – VTKT- 42 Phan Nguyen Hoai Nam (2019), “The travel demand allocation model and its practice in Vietnam”, Transport and Communication Journal, December 2019 issue, pp 143-147 Phan Nguyen Hoai Nam (2020), “Determining factors that effect the transportation mode choice behaviors in Ho Chi Minh city”, Transport and Communication Journal, August 2020 issue, pp 147-150 ... three points of view, namely: a psychological point of view, a marketing point of view and an economic point of view 2.1.1 The behavior of the person making the trip from the psychological point of... to analyze the trip choice of the trip maker There are two approaches to modeling urban travel demand, namely: the direct approach and the sequential (indirect) structural choice model approach... the psychological, marketing and economic points of view, and at the same time clarifies the theoretical basis of the process of forecasting travel demand in general and forecasting choice choose