1. Transportation Planning Asian Institute of Technology
Modal Split Model 2
Transportation Planning
Asian Institute of Technology

2. Contents Trip Interchange Modal Split Model Generalized Cost
Logit Model
Choice Model
Utility

3. Trip-interchange Modal Split Model
Splits trip interchanges into modes
Uses transportation mode characteristics and socioeconomic characteristics to estimate mode share.
Suitable for areas with high quality and competitive public transportation modes.
Applied after trip distribution model.
Advantage – Sensitive to transportation policy
Disadvantage – Data intensive. Variables are not constant over time.

4. Trip-interchange Modal Split Model
Assumptions
Public transport users are captive riders
Choice riders may choose car or public transport depending upon generalized costs comparison
where Vijc = Travelers from i to j who use car
Vijch = Travelers from i to j who are choice riders
pijc = Proportion of travelers from i to j who choose car Vijb = Travelers from i to j who use bus
pijb = Proportion of travelers from i to j who choose bus
Vijca = Travelers from i to j who are captive riders

5. Generalized Cost Monetary Costs Time-related Costs
Fuel
Fare
Toll
Parking Fee
Time-related Costs
In-vehicle travel time
Wait time
Transfer/access time
Other forms of costs
Convenience – seats, walking distance, connection
Safety
Punctuality

6. Generalized Cost
All types of costs should be converted into a common unit
Value of time converts time into monetary cost
Ex. Mr. A’s value of time is 100 baht per hour. He spent 30 minutes traveling on public transport and pays 20 baht fare. Thus, disregarding other immeasurable costs, he has a generalized costs of
Convenience, safety and other associated costs require detailed analysis to convert into monetary value. They sometimes are included in value of time of the mode.
Ex. Connection between train and taxi may be equivalent to 20 baht per one connection. Motorcycle taxi may costs 10 baht/kilometer more on safety.

7. Generalized Cost Example
Mr. A spent 5 minutes walking to a bus stop. He spent 10 minutes waiting before he could get on the bus for which he paid 20 baht. 30 minutes later he got off and continued walking for another 5 minutes to the train station. He waited for 5 minutes there and got on the train, paying 20 baht. The train took him 15 minutes to the destination station from which he walked for 5 minutes to meet his friend.
If Mr. A’s values of time for these activities are as follows:
In-vehicle time 90 baht per hour
Walk time 90 baht per hour
Wait time 120 baht per hour
Transfer 30 baht per transfer
Find Mr. A’s generalized cost for his trip to see his friend.

8. Generalized Cost

9. Multinomial Logit Model
Aggregate Level Modal Split Model
Mode share can be calculated from a logit model
where Pijm = proportion of travelers from i to j who choose mode m
Zijm = generalized cost for trip from i to j by mode m
considering only time and out-of-pocket expenses
where TTijm = travel time on trip from i to j by mode m
TCijm = out-of-pocket expenses on trip from i to j by mode m
v = value of time

10. Multinomial Logit Model
Effect of b to mode share
Dominating mode share

11. Binomial Logit Model (BNL)
Special case in which only two modes are available
Often used to analyze competition between public and private modes

12. Trip-interchange Model Application
Example A study found that skilled and unskilled labors in District A were choice and captive riders as shown
skilled
unskilled
TAZ 2 3 4 5 6 7
Vi1ch
Vi1ca
Vi8ch
Vi8ca
4020
2680
480
320
2080
1020
470
230
4427
1673
2323
877
15400
3400
16200
6400
20100
9900
Vi1ch
Vi1ca
Vi8ch
Vi8ca
7500
17500
1500
3500
4980
11620
1770
4130
8318
5582
2932
1968
3300
2200
1200
800
2098
202
1277
123
3780
420
2970
330

13. Trip-interchange Model Application
We estimated value of time for 80 baht per hour for skilled workers and 50 baht per hour for unskilled workers. The following shows travel time and cost for trips among the zones.
Expense (baht)
Travel time (minute)
TAZ
To Zone 1
To Zone 8
Car
Bus 2 15 20 40 3 35 4 25 5 30 6 7 45
To Zone 1
To Zone 8
Car
Bus 12 18 16 32 13 26 20 40 10 14 28 24

14. Trip-interchange Model Application
Example (cont’d) If binomial logit model applies with b = 0.1, find number of travelers who uses bus and car for each labor category.

15. Trip-interchange Model Application
Skilled Workers Generalized costs value of time v = 80 baht/hour = 1.33 baht/minute Zijm matrix:
Zone
To Zone 1
To Zone 8
Car
Bus 2
31.0
44.0
61.3
62.7 3
37.3
54.7
52.3 4
51.7
73.3 5
47.3
42.3 6
33.3
46.7 7
63.7
57.3
52.0

16. Trip-interchange Model Application
Skilled Workers Proportion of users and pijm matrix:
Zone
To Zone 1
To Zone 8
Car
Bus 2
0.786
0.214
0.533
0.467 3
0.850
0.150
0.558
0.442 4
0.897
0.103 5
0.676
0.324
0.774
0.226 6
0.791
0.209 7
0.347
0.653
0.891
0.109

17. Trip-interchange Model Application
Skilled Workers Car Users Bus Users
Zone
To Zone 1
To Zone 8
Car
Bus 2
3159
3541
256
544 3
1768
1332
262
438 4
3972
2128
2084
1116 5
10403
4997
2633
767 6
8639
7561
5065
1335 7
6970
13130
8820
1080

18. Logit Model
Advantages – Used to test market sensitivity to transport policy
Disadvantages – Does not distinguish between short and long distance trips.
Caution – Mode choice decision must be independent.

19. Assignment #8
Continue in-class example for unskilled workers

20. Choice Model
Takes into account individual choice before using aggregation technique to estimate mode share.
Choice model applies Utility Maximization concept
Travelers seek to minimize out-of-pocket expenses and travel time, and maximize convenience and safety.
Random utility theory – travelers may not always choose the best option available and may not choose the same alternative under the same situation, but will choose the best perceived option at a given time.

21. Utility
Utility of a traveler is a measure of his/her satisfaction to a trip by a transport mode or series of transportation modes.
Utility of mode m between i and j can be measured by its characteristics including expenses, speed (travel time), convenience, safety, and socioeconomic of the traveler.

22. Choice Model Application
Example (Meyer, M.D. and Miller, E.J.) A study revealed a choice model wth the following utility function where IVTTa and IVTTt are in-vehicle travel time for car and bus OVTTa and OVTTt are travel time outside vehicle for car and bus OPTCa and OPTCt are travel expenses for car and bus INC is traveler’s household income AO is traveler’s auto ownership

23. Choice Model Application
Example (cont’d)
Model paratmeters bn were found as follows: b1 b2 b3 b4 b5 b6
0.25
-0.1
-0.11
-0.2

24. Choice Model Application
Interview with ten sampled individuals yield the following data:
Obs #
IVTTa
IVTTt
OVTTa
OVTTt
OPTCa
OPTCt
INC AO 1 20 25 5 10
250 50 2 35 15
300 3 18 8
225 4 30 40
400 6 12
150 7
100
600 75 9
450
125

25. Choice Model Application
Calculate utilities for car and bus
Probabilities of using car and bus
Obs # Ua Ut Pa 1
-2.675
-4.785 2
-2.820
-6.862 3
-1.429
-3.598 4
-3.593
-7.412 5
-2.605
-5.318 6
-1.205
-2.355 7
-2.385
-3.768 8
-3.890
-6.418 9
-3.658 10
-1.588
-2.685

26. Aggregation of Choice Model
Three aggregation methods
Enumeration: find average of probabilities
Naïve aggregation: find average of each variable and use them to calculate probabilities.
Classification: categorize travelers by socioeconomic characteristics

27. Enumeration Obs # Ua Ut Pa 1 -2.675 -4.785 0.89187 2 -2.820 -6.862 3
-1.429
-3.598 4
-3.593
-7.412 5
-2.605
-5.318 6
-1.205
-2.355 7
-2.385
-3.768 8
-3.890
-6.418 9
-3.658 10
-1.588
-2.685
Average

28. Naïve Aggregation Obs # IVTTa IVTTt OVTTa OVTTt OPTCa OPTCt INC AO UaUt Pa 1 20 25 5 10
250 50 2 35 15
300 3 18 8
225 4 30 40
400 6 12
150 7
100
600 75 9
450
125
Avg
21.0
28.0
4.9
9.3
290.0
52.5
2.0
1.4
-2.582
-4.958

29. Classification Obs # INC AO IVTTa IVTTt OVTTa OVTTt OPTCa OPTCt Ua UtPa 1 20 25 5 10
250 50 6 12 3
150 15
125
Avg
13.3
17.3
4.3
6.7
175.0
50.0
-1.823
-3.275
Obs #
INC AO
IVTTa
IVTTt
OVTTa
OVTTt
OPTCa
OPTCt Ua Ut Pa 5 2 1 20 30 10
300 50 9 40
450
Avg
25.0
35.0
5.0
10.0
375.0
50.0
-3.131
-5.868

30. Classification Obs # INC AO IVTTa IVTTt OVTTa OVTTt OPTCa OPTCt Ua UtPa 3 2 15 18 8
225 50 7 25 5
100
Avg
15.0
21.5
5.5
6.5
162.5
50.0
-1.907
-3.683
Obs #
INC AO
IVTTa
IVTTt
OVTTa
OVTTt
OPTCa
OPTCt Ua Ut Pa 4 3 1 30 40 5 15
400 50
Avg
30.0
40.0
5.0
15.0
400.0
50.0
-3.593
-7.412
Obs #
INC AO
IVTTa
IVTTt
OVTTa
OVTTt
OPTCa
OPTCt Ua Ut Pa 2 3 25 35 5 15
300 50 8 40 10
600 75
Avg
30.0
37.5
5.0
12.5
450.0
62.5
-3.355
-6.640