The previous indicators are computed for each element of each class and are associated to each class representative as the ensemble average computed over all the class elements.
The average is computed over three sets of test runs.
The averaging was carried out not once, but repeatedly, using the algorithm (1) f n (k ) = 1 2 n + 1 ∑ k = − n n f n − 1 (k ), n = 1, 2, …, 5, f 0 (k ) = f (k ), where every new averaging was performed on the previous function f n −1(k) over a window with a greater width d = 2 n + 1; thus, the first averaging was over three elements, the second one was over five elements, and so on.
This first calculation shows that a better correlation is achieved when considering an average secular variation computed over a five year time period (SV2) before the reference epoch, than when considering an average secular variation computed over a significantly longer time period (SV1).
The average throughput is computed over the five days, including the periods of time where the battery is fully discharged (in this case the data-rate is null).
Due to the high interdependency of within population estimates as revealed by the high genetic correlations, the above overall average of estimates were computed over the two populations averages rather than over the estimates for the five individual clutches of the two populations.
These slopes are averaged for a best estimate of the slope, and the standard error computed over these two measurements is also reported.
Monthly averages were computed over the previous 10 days period of transport (counting backwards).
On average, there are approximately three elements per query.
The proposed UbiSOM algorithm relies on two learning assessment metrics, namely the average quantization error and the average neuron utility, computed over a sliding window.
For each feature we compute its average biserial correlation (over five correlations computed from the five training subsets of the MxD dataset) with the outcomes, i.e. annotation of ordered and disorder residues.
