Scholarly record

ERROR MODEL IN MODELLING AND ESTIMATIG MOVEMENT OF CYCLISTS

Kamil Peterek, Assoc. Prof. Dr. Ales Ruda, Ludmila Flokova, Dr. Iveta Dockalikova

Abstract

Using interpolation techniques for predicting economic activities such as a movement of cyclist is always vitiated by an error. Therefore, this paper deals with the issue of understanding the error during modelling and compares different assessing criteria for choosing the most accurate method in order to decrease error propagation. The diagnostic of interpolation methods can be calculated with cross-validation approach. In case of kriging methods, we computed the estimated prediction standard errors and we examined the variability of predictions from true values. In detail, we compared mean error, root mean square error, average standard error and root mean square standardized error. In addition to error indicators calculations, the fictitious error surface was created in order to facilitate orientation in estimated values loaded with high calculation error. The cross-validation showed that IDW method and Radial Base Function have worst results although IDW was the most accurate in prediction of furthest point. Opposite, Empirical Bayesian Kriging and Ordinary Kriging (spherical variogram) achieved very similar values bringing best predicts. Though kriging is very accurate interpolator, the behavior of cyclists is determined by many other factors which cannot be completely included during kriging.

Publication Impact Profile

• Captures
• Mendeley - Readers: 1

PLUMX see details

DOI resolver: doi.org/10.5593/sgem2017/21/s08.111

Publication details

Back to publication list