Federated database systems use wrappers to translate languages that may differ between database management systems.
Although variable selection using wrapper strategies may incur extra computational costs, this is compensated by the ability to explore complex associations between variables detected within the intrinsic patterns incorporated in the discrimination rules.
The improvement obtained using wrapper FSS was not as significant as when using other supervised algorithms.
SVMs have also been employed for feature selection using Wrapper and Embedded methodologies.
The number of features selected using wrapper FSS were similar but the main difference was in the searching technique.
For selecting the best features, the following methods were used: Wrapper (Kohavi and John, 1997), Relief Attribute Evaluation with Ranker (Witten and Frank, 2005) and PCA (Pearson, 1901).
Therefore, an appropriate selection of variables (using wrapper FSS in our case) appears to be more important than using a specific supervised algorithm.
Statistical hypothesis test outcomes confirm that models obtained with the wrapper approach are the most accurate to classify interneurons and pyramidal cells, since nine of the selected models in Table 7 are built using wrapper FSS.
Compared with the single best predictor (L-PPI) in the left wrist, models using Wrapper: NB and Wrapper: LR exhibited a slight improvement in AUCs of 0.043 for NB and 0.045 for LR.
The second uses wrapper algorithms (see Section 2.7) based on stochastic search methods such as GP for identifying optimal combinations of SNPs that are associated with disease end points.
For example, the model built using Wrapper: LR for the left wrist included 5 physical factors: L-PPI, L-Sum-30p, L-Sum-30p-PPI, L-Asys-HR75, and L-PSD-0-13Hz.
