fp_select_k_features {NeuroDecodeR}

A feature preprocessor (FP) that reduces data to the k most selective features

Description

This feature preprocessor object applies an ANOVA to the training data to find the p-value of all features. It then either uses the top k features with the smallest p-values, or it removes the features with the smallest k p-values. Additionally, this function can be used to remove the top k p-values and then use only the following j next smallest p-values (for example, this can be useful if one is interesting in comparing the performance using the most selective 10 neurons to using the next 10 most selective neurons, etc.).

Usage

fp_select_k_features(
  ndr_container_or_object = NULL,
  num_sites_to_use = NA,
  num_sites_to_exclude = NA
)

Arguments

ndr_container_or_object

The purpose of this argument is to make the constructor of the fp_select_k_features feature preprocessor work with the pipe (|>) operator. This argument should almost never be directly set by the user to anything other than NULL. If this is set to the default value of NULL, then the constructor will return a fp_select_k_features object. If this is set to an ndr container, then a fp_select_k_features object will be added to the container and the container will be returned. If this argument is set to another ndr object, then both that ndr object as well as a new fp_select_k_features object will be added to a new container and the container will be returned.

num_sites_to_use

The number of features with the smallest p-values to use.

num_sites_to_exclude

The number of features with the smallest p-values that should be excluded.

Value

This constructor creates an NDR feature preprocessor object with the class fp_select_k_features. Like all NDR feature preprocessor objects, this feature preprocessor will be used by the cross-validator to pre-process the training and test data sets.

See Also

Other feature_preprocessor: fp_zscore()

Examples

# This will cause the cross-validator use only the 50 most selective sites
fp <- fp_select_k_features(num_sites_to_use = 50)

# This will cause the cross-validator to remove the 20 most selective sites
fp <- fp_select_k_features(num_sites_to_exclude = 20)

# This will cause the cross-validator to remove the 20 most selective sites
# and then use only the 50 most selective sites that remain after the 20 are
# eliminated
fp <- fp_select_k_features(num_sites_to_use = 50, num_sites_to_exclude = 20)