Considering the result of a sport event regarding years as a time serial, it can be described by a model of time serial including certain model and uncertain model. Prediction models are aroused with the help of logarithmic, exponential, triangular models and their derivations, by which variant optimal models for score prediction of different sports are obtained. The athletic scores of Olympic in 2004 are predicted in the paper, the ratio of prediction is over 96% (97% by average). Finally, scores are predicted and post-error are tested for all athletic items according to variant models through computer simulation. Results show that logarithmic derivation model is optimal for men items and sinusoidal model is more credible for woman items, which reflects the basic rules of score chance by reference of time variance.
© Copyright 2008 Proceedings of First Joint International Pre-Olympic Conference of Sports Science and Sports Engineering. Volume I: Computer Science in Sports.. Published by World Academic Union. All rights reserved.