peal.teachers.virelay_teacher.meta_analysis¶
- peal.teachers.virelay_teacher.meta_analysis(base_dir, attribution_database_file_path, analysis_file_path, variant, class_indices, label_map_file_path, number_of_eigenvalues, number_of_clusters_list, number_of_neighbors)[source]¶
Performs a meta-analysis over the specified attribution data and writes the results into an analysis database.
- Parameters:
base_dir (
str) – The base path to the CoRelAy project.attribution_database_file_path (
str) – The path to the attribution database file, that contains the attributions for which the meta-analysis is to be performed.analysis_file_path (
str) – The path to the analysis database file, into which the results of the meta-analysis are to be written.variant (
str) – The meta-analysis variant that is to be performed. Can be one of “absspectral”, “spectral”, “fullspectral”, or “histogram”.class_indices (
List[int]) – The indices of the classes for which the meta-analysis is to be performed. If not specified, then the meta-analysis is performed for all classes.label_map_file_path (
str) – The path to the label map file, which contains a mapping between the class indices and their corresponding names and WordNet IDs.number_of_eigenvalues (
int) – The number of eigenvalues of the eigenvalue decomposition.number_of_clusters_list (
List[int]) – A list that can contain multiple numbers of clusters. For each number of clusters in this list, all clustering methods and the meta-analysis are performed.number_of_neighbors (
int) – The number of neighbors that are to be considered in the k-nearest neighbor clustering algorithm.
- Return type:
None