Calculations of standard or relative error are common ways to look at the data and departures from deviations in the data will directly affect robustness.
This average is fairly resistant to large deviations in the data; however, it is too insensitive to significant changes in average deviation to be useful as a detection technique by itself.
These differences were statistically significant only because of the extremely small standard deviations in the data.
In this experiment, the load to failure was the better factor because it showed lower standard deviations in the data.
However, we interpret this apparent change in K to be due to deviations in the data itself (such as the appearance of multiple-component or anomalous diffusion).
30 } ; M = 20 M = { 30, 20, 15, 10, 5 } ; ξ = 0. 15 Since standard deviations in the data distributions were also used as optimization weights, a small amount of absolute error (η=0.001) was added to ensure errors in small values did not dominate the cost function.
