The phrase "average model size" is correct and usable in written English.
It can be used when discussing the typical dimensions or parameters of a model, often in contexts like machine learning, statistics, or product specifications.
Example: "The average model size for our dataset is approximately 150 megabytes, which allows for efficient processing."
Alternatives: "typical model size" or "mean model size".
SPCA tends to select a larger model as represented by its average model size and high FPR.
SPCA has high detection rates for all non-zero coefficients, but the use of this method is restricted by its considerably large FPR and average model size.
Average estimated effects, empirical standard errors, percentages of correct identification of non-zero coefficients, and average model size corresponding to four statistical methods in a time-series study with count response and 4 air pollutants.
Average estimated effects, empirical standard errors, percentages of correct identification of non-zero coefficients, and average model size corresponding to four statistical approaches in a time-series study with count response and 10 air pollutants.
Average estimated effects, empirical standard errors, percentage of correct identification of non-zero coefficients, and average model size corresponding to 5 statistical methods in a cross-sectional study with continuous responses and 4 candidate air pollutants.
For each examined study design, we present the average of estimated effects with their empirical standard errors for non-zero predictors, the percentage of models correctly identifying the non-zero coefficients, and the average model size over 1000 replicates for five different statistical methods (BMA, DSA, LASSO, PLSR and SPCA).
This is reinforced by the fact that the average model sizes do not change significantly between different runs (cf. Supplementary Material).
In terms of variable selection, BMA and LASSO regression are preferred with lower FPRs, high TPRs, and average model sizes close to the true one.
Compared to the simulation results in Table 2(A), there are substantial improvements on the variable selection for all approaches, as reflected by the appreciable reduction in average fitted model size and FPR.
Open image in new window Fig. 10 Linear regression: group size average (model observation data).
Overall, the average change in gene model size was 27.9%, a substantial general increase in the sequence length of the Myc-GeneModel set, although not necessarily a change in the translated sequence.
