Scribd
Upload
32 views6 pages

Lecture 06 - Transportation Demand Analysis

Transportation demand modeling is essential for predicting travel demand and guiding planners in decision-making, utilizing Origin-Destination matrices and zonal analysis. Forecasting techniques include time series models, moving averages, and regression models, which help estimate trips and account for trends and variations. Methods like exponential smoothing and weighted averages enhance accuracy in predictions by assigning different importance to historical data.

Uploaded by

Kne 2434
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
32 views6 pages

Lecture 06 - Transportation Demand Analysis

Transportation demand modeling is essential for predicting travel demand and guiding planners in decision-making, utilizing Origin-Destination matrices and zonal analysis. Forecasting techniques include time series models, moving averages, and regression models, which help estimate trips and account for trends and variations. Methods like exponential smoothing and weighted averages enhance accuracy in predictions by assigning different importance to historical data.

Uploaded by

Kne 2434
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Lecture 06: Transportation Demand Analysis

I. Transportation Demand Modeling


Transportation demand modeling is crucial for predicting travel demand and aiding
planners in decision-making. Analysts develop accurate and user-friendly models to estimate
trips between urban zones, traffic loads, and passenger kilometers. Since individuals choose
their travel mode, time, and route based on personal factors, predicting behavior is complex.
Due to data limitations and high costs, transportation analysis is typically conducted at a zonal
level rather than an individual level. Regions are divided into transportation analysis zones,
each represented by a centroid that generates (origin) or attracts (destination) traffic flows.
These models help optimize transportation planning and infrastructure development.

The Origin-Destination (O-D) matrix records traffic flow between centroids, helping
transportation planners analyze and predict travel demand. Short-term O-D matrices are often
unbalanced due to time-based travel variations, but they balance out over a full day. Planners
estimate and forecast trips between zones based on factors such as household size, car ownership,
income, and road density.

Macroscopic models predict transportation trends at a regional or national level,


estimating trips, passenger kilometers, and aircraft operations. Microscopic models focus on
specific routes, modes, or fare classes, helping optimize transportation planning and operations.

II. Transportation Demand Forecasting Techniques


Transportation demand forecasting accounts for trends, cyclical patterns, seasonal peaks,
and random variations. Trends indicate long-term growth or decline, while seasonal and
cyclical components reflect recurring fluctuations. Random variations result from
unpredictable events and are usually short-term.

Forecasting relies on historical data, but all predictions contain errors, with shorter-term
forecasts being more accurate. Qualitative methods, such as the Delphi method, rely on expert
judgment when data is limited. Quantitative techniques include extrapolation (time series
models) and explanatory (regression models). Extrapolation forecasts demand based on past
trends, while regression models predict demand using socioeconomic and transportation
factors, allowing "what-if" scenario analysis. Time series methods include the naive approach,
moving averages, and trend projection.

Figure 1. Forecasting Techniques

a. Time Series Model


a.1. Naive Approach
The naive approach is based on the assumption that demand in the next time
period is identical to demand in most recent time period.
Example: If today’s ridership is 50,000 passengers, the naive forecast for tomorrow is
also 50,000 passengers.
Formula:
Forecast t+1 = Actual t

a.2. Moving Average Method


The moving average model utilizes previous t periods in order to predict demand
in the time period t+1. Analysts use simple moving average and weighted moving
average method. The forecast value in the simple moving average method equals:

Example 1:
A transportation agency wants to predict the number of daily passengers on a
subway system. They decide to use a 5-day moving average to smooth out short-
term fluctuations and identify trends. Calculate the moving average for day 6.

Given Data (Past 5 Days Passenger Counts):


Day Passenger Count
1 50,000
2 52,000
3 51,500
4 53,000
5 54,500

Example 2: The data on passenger traffic at Belgrade Airport, Serbia, from 1970 to
1977 are given. Predict the number of passengers in 1975 by using a 3-year moving
average.

Year Passenger Traffic at Belgrade Airport


1970 838,156
1971 1,036,311
1972 1,155,166
1973 1,434,454
1974 1,688,247
1975 2,020,291
1976 2,047,016
1977 2,280,972
a.3. Weighted Moving Average Method
The weighted moving average method allows analyst to give more importance
to some of the historical data. The forecast value in the weighted moving average
method equals:

Example 1: A city's transportation department wants to predict subway ridership for


the next day using a 3-day weighted moving average. They assign higher weights to
more recent days because recent trends impact future ridership more.

Given Data (Past 3 Days Passenger Counts):


Day Passenger Count Weight
1 50,000 0.2
2 52,000 0.3
3 51,500 0.5

Example 2: Predict the number of passengers in 1975 at Belgrade Airport, by using


the weighted moving average method. Use a 3-year moving average, and the
following set of weights:

W1974 = 0.6
W1973 = 0.3
W1972 = 0.1
Year Passenger Traffic at Belgrade Airport
1970 838,156
1971 1,036,311
1972 1,155,166
1973 1,434,454
1974 1,688,247
1975 2,020,291
1976 2,047,016
1977 2,280,972
a.4. Exponential Smoothing
Exponential smoothing is based on the idea of assigning higher weights to more
recent observations than to the observations from the far-away past. In other words,
in the case of exponential smoothing, the weights decrease exponentially with the
past. In this way, the lowest weights are given to the most distant observations. The
forecast value in the exponential smoothing equals:

Example 1: Assume we are forecasting daily traffic volume using simple exponential
smoothing.
Given: At = 10,000 vehicles today. Previous forecast is 9,500 vehicles.
Smoothing factor is 0.3.

Example 2: Predict the number of passengers at Belgrade Airport, by using the


exponential smoothing method. Use the smoothing constant 𝛼 = 0.8. Assume that F1
= A1 .
Year Passenger Traffic at Belgrade Airport
1970 838,156
1971 1,036,311
1972 1,155,166
1973 1,434,454
1974 1,688,247
1975 2,020,291
1976 2,047,016
1977 2,280,972

Seatwork: A city’s transportation department wants to analyze annual train ridership


trends using a 5-year moving average to smooth out fluctuations and identify long-
term patterns.

Given Data (Annual Train Ridership in Millions):


Year Passenger Count (millions)
2015 300
2016 320
2017 340
2018 360
2019 380
2020 400
2021 420
2022 440
2023 460

a. Compute the 5-year moving averages.


b. Predict the number of ridership using exponential smoothing method. Use the
smoothing constant 𝛼 = 0.7. Assume that F1 = A1.

You might also like