provides software and
services for engineering, science, and business modeling and simulation
applications.
Its major product is TaylorFit
Software for empirical modeling of complex systems. TaylorFit
is a software implementation that helps a user to develop Multivariate
Polynomial Regression (MPR) models.
- Do linear models leave you flat?
- Do artificial neural networks make you nervous?
- Are you in search of a higher correlation?
MPR is new class of models for use in forecasting and correlation
analysis. An MPR model is essentially a multiple regression model
with polynomial and cross-product (interaction) terms. For example,
if Y is a function of Q, R, and S, terms can be included such as
QR2S or Q3S.
MPR models can be fitted using conventional multiple regression
methods, and only terms that are statistically significant are retained
in the model. MPR models are applicable to low-to-moderate dimensionality
problems as are encountered in science, engineering and business.
|
MPR models can replace linear modeling techniques such as multilinear
regression and ARMA (Box-Jenkins) models, and can produce far superior
results because they incorporate nonlinear effects. If you have
used one of these linear methods in the past, you should dust off
your data and take another look with TaylorFit. Otherwise you could
be missing important behaviors in your data.
MPR models compare favorably to artificial neural network (ANN)
models: MPR models can provide a better fit with fewer coefficients;
it is easier to control overfitting, or "memorizing" of
data; the fitting procedure is less computationally intensive and
converges absolutely; MPR models do not require a priori selection
of model structure; they give a simple explicit equation for prediction
or analysis; standard statistical tests can be applied to all coefficients
and forecast predictions.
MPR models can also be used in most applications to which artificial
neural networks have been applied. However, MPR models have a number
of advantages, especially that they are easier to understand and
use, and they make it easier to avoid overfitting errors.
|