Model Calibration and Validation
The following excerpt was taken from the Transportation Planning Handbook
published in 1992 by the Institute of Transportation Engineers (p. 116).
(p.116) The process of developing travel models is commonly called
"calibration." Given the basic form of a travel forecasting model, such as a
gravity model or a logit model, calibration involves estimating the values of various
constants and parameters in the model structure. For this reason the model development
effort is sometimes termed "estimation."
Estimating model coefficients and constants is usually done by solving the model
equation for the parameters of interest after supplying observed values of both the
dependent and independent variables. The observed values of variables are obtained from
the surveys of actual travel patterns. As indicated previously, the estimation process is
a trial and error effort that seeks the parameter values which have the greatest
probability or maximum likelihood of being accurate within acceptable tolerance of error.
Such an effort is commonly accomplished with specialized statistical computer programs
designed for just such purposes. . . . Model calibration can also be accomplished by using
values of constants and parameters from models estimated for another location that is
similar to the area being studied; this strategy is referred to as "importing"
model parameters and should be employed only by experienced practitioners.
Once satisfactory estimates of the parameters for all models have been obtained, the
models must be checked to assure that they adequately perform the functions for which they
are intended, that is, to accurately estimate traffic volumes on transit and roadways.
Verifying a calibrated model in this manner is commonly called "validation." The
validation process establishes the credibility of the model by demonstrating its ability
to replicate actual traffic patterns.
Validating the models requires comparing traffic estimated by the model to observed
traffic on the roadway and transit systems. Initial comparisons are for trip interchanges
between quadrants, sectors, or other large areas of interest. . . . The next step is to
compare traffic estimated by the models to traffic counts, including transit ridership,
crossing contrived barriers in the study area. These are commonly called screenlines,
cutlines, and cordon lines and may be imaginary or actual physical barriers. Cordon lines
surround particular areas such as the central business district or other major activity
centers. . . . Transit ridership estimates are commonly validated by comparing them to
actual patronage crossing cordon lines around the central business district. . . .
The importance of traffic and transit counts for model validation underscores the need
for careful planning, thoroughness and accuracy of a traffic and transit data collection
program that has this purpose. As with the travel surveys, the resulting models and
forecasts will be no better than the data used for model estimation and validation.
|