The phrase "a proportion of features" is correct and usable in written English.
It can be used when discussing the relative amount or percentage of certain characteristics or attributes within a larger set.
Example: "In this study, we found that a proportion of features were significantly correlated with the outcomes we measured."
Alternatives: "a percentage of features" or "a fraction of features".
Two models of item similarity were tested by assuming that the item representations share a proportion of features and that the exemplars from different stimulus classes vary in the distinctiveness or diagnosticity.
It is a backward feature selection approach which starts with the entire feature set and iteratively removes a proportion of features after evaluation using SVM, which was implemented using a toolbox in Matlab 8.3 [30].
The algorithm begins by subsampling a proportion of items and a proportion of features from a data matrix.
The aim of this stage was to increase the level of consistency across participants to allow as high a proportion of features to be automatically collated as possible.
Unlike the results from the individual sparse k-means method, the number of features contributing to class discrimination was restricted for all three integrative analyses (i.e. sparsity was enforced and a proportion of features were assigned weights of zero), with the Blood-Cranial clustering analysis having the fewest number of contributing features.
Here, it is assumed that stimuli from the same category share a proportion of their features in common with each other and that members of different categories have far fewer features in common.
In sparse k-means clustering each feature is given a non-negative weight and depending on the magnitude of the tuning parameter, a proportion of the features will be given a weight of zero indicating that they do not contribute to the clustering (i.e. sparsity was enforced).
This scenario would violate the assumption that only a small proportion of features are expected to show differential expression in a microarray dataset.
Figures 14(b) and 14(c) shows that after LS pairing, a large proportion of feature pairs have relatively moderate "size" densities and moderate values.
Some of these proposals would ban or restrict a proportion of Uber's key features.
This indicates that there is a proportion of pores with physical features that differ considerably from those of the majority present in the piassava fiber.
