Common use of Model Verification Clause in Contracts

Model Verification. The best performing models from the comprehensive modeling process (those with the highest R2 values) were selected and investigated further. Models with collinearity among the predictor variables were removed from consideration. Often, multiple models have similar performance, so the logic for choosing one comes down to factors such as explainability and implementability. A simpler model, one with fewer predictor variables or fewer interaction terms, is often easier to understand and explain, and easier to use. It might be better to choose a model because it is easier to explain and use even if it performs slightly less well (i.e. no practical difference). Models that tend to tie to tree physiology and simple parameters are easy to characterize and interpret. The remaining model with the highest R2 value was then examined looking for non-significant coefficients. If interaction terms were non-significant, they were removed. The model was plotted, again looking for undesirable behavior, for example, significant trending captured by sloping in the residuals.