خوشه بندی روی داده buddymove holidayiq

clc;
clear;
close all;

% Amir Shokri
% amirsh.nll@gmail.com
% github.com/amirshnll
% July 2020

% Load Data
dataFile = importdata('buddymove_holidayiq.csv');
data = dataFile.data;
textdata = dataFile.textdata;
colheaders = dataFile.colheaders;

% Data Normalized
data_normal = normalize(data,'range');

% TTSAS Algorithm 
theta1 = 2.2; 
theta2 = 4;
TTSAS_labels = TTSAS(data,theta1,theta2);
figure, scatter( data(:,1), data(:,2), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,1), data(:,3), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,1), data(:,4), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,1), data(:,5), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,1), data(:,6), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,2), data(:,3), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,2), data(:,4), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,2), data(:,5), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,2), data(:,6), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,3), data(:,4), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,3), data(:,5), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,3), data(:,6), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,4), data(:,5), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,4), data(:,6), 5, TTSAS_labels, 'filled' );
figure, scatter( data(:,5), data(:,6), 5, TTSAS_labels, 'filled' );

function rep = getClusterRepresentative(inds, X)

% Amir Shokri
% amirsh.nll@gmail.com
% github.com/amirshnll
% July 2020

rep = mean( X(inds,:), 1 )';
end

function [d_x_i_C_k,k] = findClosestCluster( ii, labels, X )
% Amir Shokri
% amirsh.nll@gmail.com
% github.com/amirshnll
% July 2020
ulabels = unique(labels);
if( ulabels(1)==0 ) 
  ulabels = ulabels(2:end);
end
x_ii_to_cluster = [];
for lab = ulabels,
  inds = find( labels==lab );
  rep  = getClusterRepresentative( inds, X ); 
  d = sqrt( ( X(ii,:)' - rep )' * ( X(ii,:)' - rep ) ); 
  x_ii_to_cluster = [ x_ii_to_cluster, d ]; 
end
[d_x_i_C_k,mind] = min(x_ii_to_cluster); 
k = ulabels(mind);

function [thetas,numFoundClusters] = estimateNumberOfClusters(X,s,Nsteps, q)
% Amir Shokri
% amirsh.nll@gmail.com
% github.com/amirshnll
% July 2020
N         = size(X,1);
nFeatures = size(X,2);
Xd = pdist(X, 'euclidean');
a = min(Xd);
b = max(Xd);
thetas           = linspace( a, b, Nsteps );
numFoundClusters = zeros( s, Nsteps ); 
for ti = 1:Nsteps,
  t = thetas(ti); 
  for si = 1:s, 
    labs = BSAS( X(randperm(N),:), t, q ); 
    numFoundClusters(si,ti) = length(unique(labs));
  end
end

function labels = TTSAS(X,theta1,theta2)

% Amir Shokri
% amirsh.nll@gmail.com
% github.com/amirshnll
% July 2020

N         = size(X,1);
nFeatures = size(X,2);
m             = 0;
is_labeled    = zeros(N,1);
prev_change   = 0;
cur_change    = 0; 
exists_change = 0; 
labels        = zeros(1,N); 
while( sum(is_labeled)<N )
  first_in_while_loop = 1; 
  for ii=1:N
    if( is_labeled(ii)==0 && first_in_while_loop==1 && exists_change==0 )
      first_in_while_loop = 0; 
      m              = m+1; 
      labels(ii)     = m; 
      is_labeled(ii) = 1; 
      cur_change     = cur_change + 1; 
    elseif( is_labeled(ii)==0 )
      [d_x_i_C_k,k] = findClosestCluster( ii, labels, X ); 
      if( d_x_i_C_k < theta1 )
	labels(ii)     = k; 
	is_labeled(ii) = 1; 
	cur_change     = cur_change + 1; 
      elseif( d_x_i_C_k > theta2 )
	m              = m+1; 
	labels(ii)     = m; 
	is_labeled(ii) = 1; 
	cur_change     = cur_change + 1; 
      end
    elseif( is_labeled(ii)==1 )
      cur_change = cur_change + 1; 
    else
      fprintf('should never get here\n'); 
    end
  end
  exists_change = abs( cur_change - prev_change ); 
  prev_change   = cur_change; 
  cur_change    = 0;
end

clc;
clear;
close all;

% Amir Shokri
% amirsh.nll@gmail.com
% github.com/amirshnll
% July 2020

% Load Data
dataFile = importdata('buddymove_holidayiq.csv');
data = dataFile.data;
textdata = dataFile.textdata;
colheaders = dataFile.colheaders;

% Data Normalized
data_normal = normalize(data,'range');

% Fuzzy c-Means Algorithm 
options_fcm = [2 100 0.00001 false];
cluster_size = 5;
[C_fcm,U_fcm] = fcm(data,cluster_size,options_fcm);
maxU_fcm = max(U_fcm);
index_fcm = cell(cluster_size,1);
for i=1:cluster_size
    index_fcm{i} = find(U_fcm(i,:) == maxU_fcm);
end
figure
hold on
for i=1:cluster_size
    plot(dataset(index_fcm{i},1),dataset(index_fcm{i},2),'o')
    plot(C_fcm(i,1),C_fcm(i,2),'x','MarkerSize',15,'LineWidth',2)
end
hold off

clc;
clear;
close all;

% Amir Shokri
% amirsh.nll@gmail.com
% github.com/amirshnll
% July 2020

% Load Data
dataFile = importdata('buddymove_holidayiq.csv');
data = dataFile.data;
textdata = dataFile.textdata;
colheaders = dataFile.colheaders;

% Data Normalized
data_normal = normalize(data,'range');

% Min Proximity
Z = linkage(data,'ward','chebychev');
T = cluster(Z,'maxclust',3);
cutoff = median([Z(end-2,3) Z(end-1,3)]);
dendrogram(Z,'ColorThreshold',cutoff)

clc;
clear;
close all;

% Amir Shokri
% amirsh.nll@gmail.com
% github.com/amirshnll
% July 2020

% Load Data
dataFile = importdata('buddymove_holidayiq.csv');
data = dataFile.data;
textdata = dataFile.textdata;
colheaders = dataFile.colheaders;

% Data Normalized
data_normal = normalize(data,'range');

% Max Proximity
Z = linkage(data,'average','chebychev');
T = cluster(Z,'maxclust',3);
cutoff = median([Z(end-2,3) Z(end-1,3)]);
dendrogram(Z,'ColorThreshold',cutoff)