Reference API

This is the primary reference of snfpy. Please refer to the user guide for more information on how to best implement these functions in your own workflows.

snf.compute - Primary SNF functionality

Contains the primary functions for conducting similarity network fusion workflows.

make_affinity(*data[, metric, K, mu, normalize])	Constructs affinity (i.e., similarity) matrix from data
get_n_clusters(arr[, n_clusters])	Finds optimal number of clusters in arr via eigengap method
snf(*aff[, K, t, alpha])	Performs Similarity Network Fusion on aff matrices
group_predict(train, test, labels, *[, K, mu, t])	Propagates labels from train data to test data via SNF

snf.metrics - Evaluation metrics

Functions for computing various metrics to aid interpretation of similarity network fusion outputs.

nmi(labels)	Calculates normalized mutual information for all combinations of labels
rank_feature_by_nmi(inputs, W, *[, K, mu, …])	Calculates NMI of each feature in inputs with W
silhouette_score(arr, labels)	Calculates modified silhouette score from affinity matrix
affinity_zscore(arr, labels[, n_perms, seed])	Calculates z-score of silhouette (affinity) score by permutation

snf.cv - Cross-validation functions

Code for implementing cross-validation of similarity network fusion. Useful for determining the “optimal” number of clusters in a dataset within a cross-validated, data-driven framework.

snf_gridsearch(*data[, metric, mu, K, …])	Performs grid search for SNF hyperparameters mu, K, and n_clusters
get_optimal_params(zaff, labels[, neighbors])	Finds optimal parameters for SNF based on K-folds grid search

snf.datasets - Load tests datasets

Functions for loading test data setss

load_simdata()	Loads “similarity” data with two datatypes
load_digits()	Loads “digits” dataset with four datatypes