The parameter estimates from noise-free data indicated that both IFPE variants could produce nearly unbiased estimates with median parameter errors of less than 1%.
When you're away, it'll just go with the median parameters to keep your plant alive.
However, when the true model (TM) is known, the results have greater coverage of the true population parameter value, smaller median relative error, smaller median relative 95% HPD bound, and smaller absolute median relative bias than model-averaged estimates of θ.
The median minimal error rate over all parameter values (k = 1, 3, 5), all numbers of genes (p* = 20, 50, 100, 200, 500) and the three gene selection methods (t-statistic, Wilcoxon, limma) is 0.43.
Basic statistical parameters, such as minimum, maximum, mean, median, standard error (SE), and skewness, were computed along with correlation study.
First, the detector identifies zero, one, or more peaks (identified with red plus in Figure 4) based on a shoulder sensitivity parameter that we set to two times the median analytical error of all UBP regions in the larger event (all error bars above the peak threshold in Figure 4).
The median relative errors of the parameter estimates from the IFPE, IPE and SPE methods are given in Table 1.
In addition, the median standard error based on the observed information was markedly less than the standard deviation of bootstrap parameter estimates; this was particularly pronounced for the absolute sigmoid model, which was generally the best-fitting of the 2-parameter symmetric curve models.
