Het lineaire regressiemodel en de analyse van simulatie-experimenten.
Report 2000.LT.5368, Transport Technology, Logistic Engineering.
Simulation is like doing experiments using computermodels. The
computermodel is as a black box, where there are one or more input
variables and one or more output variables. The relationship between the
input and the output variables can be described with a model. In this
rapport techniques are described to calculate the properties and
parameters of these models, in case the models are linear.
In literature distinction is being made between models with constant input
variables and models with stochastic input variables. These models are of
y = b0 + b1x1 + b2x2 + ... + bkxk + e.
Here are the x1, x2, ... , xk
input variables, the
b1, b2, ... , bk
the regression coefficients and is y the output variable.
For the different models techniques are described to calculate the
regression variables, estimate the variance of the model, test hypotheses
and to construct confidence intervals for the regression coefficients.
The model with constant input variables can be used when y is a linear
function of known functions of input parameters. The model with stochastic
input variables can only be used when y is a linear function of the input
The results that the different techniques give for the model with constant
input variables and the model with stochastic input variables are almost
the same. Only in testing hypotheses the distributions of the test
For both models the assumptions are being made that the values of the
input parameters are known. The effect of not accurately measured
inputvariables on the values of the regression coefficients is
Reports on Logistic Engineering (in Dutch)
, TU Delft