simuclustfactor 0.0.2

Simultaneous Component and Clustering Models for Three-way Data: Within and Between Approaches.

python-simuclustfactor

Perform simultaneous clustering and factor decomposition in Python for three-mode datasets, a library utility.

The main use cases of the library are:

• performing tandem clustering and factor-decomposition procedures sequentially (TWCFTA).
• performing tandem factor-decomposition and clustering procedures sequentially (TWFCTA).
• performing the clustering and factor decomposition procedures simultaneously (T3Clus).
• performing factor-decomposition and clustering procedures simultaneously (3FKMeans).
• performing combined T3Clus and 3FKMeans procedures simultaneously (CT3Clus).

Installation

To install the Python library, run:

pip install simuclustfactor

You may consider installing the library only for the current user:

pip install simuclustfactor --user

Library usage

The package provides just two main modules namely,

• tandem: encapsulating TWCFTA and TWFCTA
• simultaneous: encapsulating T3Clus, TFKMeans and CT3Clus

>>> import numpy as np
>>> from simuclustfactor import tandem
>>> from simuclustfactor import simultaneous
>>> from sklearn.datasets import make_blobs

>>> I,J,K = 40,15,20  # dimensions in the full space.
>>> G,Q,R = 8,4,3  # tensor dimensions in reduced space.  
>>> X_i_jk, y = make_blobs(n_samples=I, centers=G, n_features=J*K, random_state=0)  # generate dataset

>>> twcfta = tandem.TWCFTA(random_state=0,verbose=True, n_max_iter=10).fit(X_i_jk, full_tensor_shape=(I,J,K), reduced_tensor_shape=(G,Q,R))

>>> twfcta = tandem.TWFCTA(random_state=0,verbose=True, n_max_iter=10).fit(X_i_jk, full_tensor_shape=(I,J,K), reduced_tensor_shape=(G,Q,R))

>>> t3clus = simultaneous.T3Clus(random_state=0, init='random', verbose=True, n_max_iter=10).fit(X_i_jk, full_tensor_shape=(I,J,K), reduced_tensor_shape=(G,Q,R))

>>> tfkmeans = simultaneous.TFKMeans(random_state=0, init='random', verbose=True, n_max_iter=10).fit(X_i_jk, full_tensor_shape=(I,J,K), reduced_tensor_shape=(G,Q,R))

>>> tfkmeans_1 = simultaneous.CT3Clus(random_state=0, init='random', verbose=True, n_max_iter=10).fit(X_i_jk, full_tensor_shape=(I,J,K), reduced_tensor_shape=(G,Q,R), alpha=0)

>>> t3clus_1 = simultaneous.CT3Clus(random_state=0, init='random', verbose=True, n_max_iter=10).fit(X_i_jk, full_tensor_shape=(I,J,K), reduced_tensor_shape=(G,Q,R), alpha=1)