Quantitative methods for transport management review

1
Quantitative methods for transport management
(review)
Nakorn Indra-Payoong
Maritime College, Burapha University
2

• I think think months and years
• The ninety-nine times the conclusion is false
• The hundred times I am right
• Albert Einstein

3
Reviews

• Regression, risk, vehicle routing
• Decision model, transport and logistics problems,
  network flow
• Discrete choice, time series

4
Mathematical models

• Forecasting
• Operation (OR)
• Integrated

5
Real world vs model world

• Real world complex, uncertain, random, error,
  stochastic
• Model world many assumptions required

6
Operation research (OR)

• Objective single, multiple, bi-level,
  multi-level, tradeoffs
• Constraints hard , soft
• Decision variables discrete, binary, continuous
• Parameters vs variables

7
Programme ltgt programme

• Linear programme assign tasks in a linear form
• Computer programme software, codes

8
Global vs local optimum

• Global optimum the best solution
• Local optimum the feasible solutions
• Feasible vs infeasible

9
Network flow models

• Nodes, arcs, source nodes, sink nodes
• Flow conservation (input output)

10
Node-link incident matrix

• Relationship between flow conservative and
  network structure
• Input node (positive sign)
• Output node (negative sign)
• Number of rows (nodes)
• Number of columns (links)

11
Transport and logistics problem

• Assignment problem
• Routing and scheduling problem
• Shortest path problem
• Multicommodity network flow
• Different costs (parameters)
• Different products (commodities)

12
Solution methods

• Mathematical methods simplex, lagrangian, column
  generation,
• Local search heuristics, meta-heuristics (tabu
  search, genetic algorithm,)
• Complete (exact method), incomplete (approximate
  method)

13
Model evaluation

• Uncertainty of data and representation
• Sensitivity of solution and model
• Robustness

14
Discrete choice model

• Discrete choice vs regression
• Logit, probit, binary logit, multinomial logit,
  nested logit, mixed logit
• Random utility theory

15
Applications

• Policy testing
• Marketing research
• Traffic (demand) diversion
• Tradeoff , value of time

16
Utility function

• People select alternative with the highest
  utility
• Utility has no unit
• Utility function composes of systematic utility
  and error (disturbance) term

17
SP and RP

• Stated preference (SP) data is collected in a
  hypothetical basis (test the market when non
  exist)
• Reveal preference (RP) data is collected in an
  actual market

18
Choice experiments

• Sampling technique
• Influencing factors
• Market segmentation
• Variable testing
• Disaggregate vs aggregate prediction

19
Model evaluation

• Evaluation techniques
• Sensitivity and robustness
• Variable forecasting

20
Regression and time series

• Example
• Dependent variable (Y) revenues of a utility
• Explanatory variable (X) price index, interest
  rate, unemployment rate, aggregate demand

21
Time series

• Productivity would be another explanatory
  variable, as would be dummy variables for time
  periods containing different rules governing the
  utility
• Present is influenced by the past and
  disturbance (error) term
• Change the number of sample N to T

22
Time series

• Time series concerns about (serial) correlation
  amongst the disturbances of a regression model
• Called autoregressive model

23
My questions

• Three questions (100)
• First question Keywords to be explained (25)
• The second a challenging transport problem is
  given, you are expected to give a conceptual
  design for the quantitative analysis (analyse
  input data, set the scope of study and
  assumptions, select the correct and appropriate
  methods, give a well-integrated framework, design
  the evaluation techniques, and discuss (40)
• The last take home exam (35)