Potts Complete Shrinkage

Potts Clustering with Complete Shrinkage

Installation

Install using pip pip install pottscompleteshrinkage

Requirements

• Python 3.6 or greater
• numpy
• pandas

Usage

Import the Potts Complete Shrinkage module

import pottsshrinkage.completeshrinkage as PCS

Choose the number of colors

q = 20

Compute Initial Potts Clusters as a first Random Partition (with Potts Model)

InitialPottsClusters = PCS.InitialPottsConfiguration(Train_PottsData_demo, q, Kernel='Mercer')

Choose your temperature (T) level

T = 1000

Set the bandwidth of the model

sigma = 1

Set the Number of Random_Partitions you want to simulate

Number_of_Random_Partitions = 3

Set your initial (random) Potts partition as computed above

Initial_Partition = InitialPottsClusters

Set the Minimum Size desired for each partition generated

MinClusterSize = 5

Run your Potts Complete Shrinkage Model to simulate the Randomly Shrunk Potts Partitions. Partitions_Sets is a dictionary that can be saved with pickle package.

Partitions_Sets,Spin_Configuration_Sets = PCS.Potts_Random_Partition (Train_PottsData_demo, T, sigma, Number_of_Random_Partitions, MinClusterSize, Initial_Partition, Kernel='Mercer')

Pypi Project Page

https://pypi.org/project/pottscompleteshrinkage/1.0.0/

Execution Code Pipeline in Jupyter Notebook

https://github.com/kgalahassa/pottscompleteshrinkage-notebook