SERA Toolbox1 and Toolbox2 standalone versions
This commit is contained in:
@@ -0,0 +1,292 @@
|
||||
% PROGRAM: Clustering
|
||||
% VERSION: [Interactive Standalone Version] V1.8
|
||||
% COMPATIBLE with Matlab version 2017b or later
|
||||
% TOOLBOX: "Clustering/Transformation to ED Toolbox" within SERA Project
|
||||
% DOCUMENT: "READ_ME_App_1B_v1_Description_Cluster_Analysis.docx"
|
||||
% -------------------------------------------------------------------------------------------------------
|
||||
% The function gathers several different clustering algorithms included in
|
||||
% MATLAB in order to perform Cluster Analysis for datasets tranformed
|
||||
% to Equivalent Dimensions
|
||||
% --------------------------------------------------------------------------------------------------------
|
||||
% INPUT:
|
||||
% THE PROGRAM USES AS INPUT THE OUTPUT DATA OBTAINED
|
||||
% AFTER RUNNING THE "ED_ToolBox_Wrapper.mat" PROGRAM
|
||||
% --------------------------------------------------------------------------------------------------------
|
||||
% OVERVIEW: This Application is a Matlab function which akes as input
|
||||
% the output file created after "ED_ToolBox_Wrapper.mat", therefore all
|
||||
% analyses are performed within the Equivalent Dimension phase space.
|
||||
% Output results (matlab structures/variables) are produced as well as
|
||||
% some figures in some particular cases.
|
||||
% --------------------------------------------------------------------------------------------------------
|
||||
% AUTHORS: K. Leptokaropoulos,
|
||||
% last updated: 03/2019, within SERA PROJECT, EU Horizon 2020 R&I
|
||||
% programme under grant agreement No.730900
|
||||
% CURRENT VERSION: v1.8 **** [INTERACTIVE STANDALONE VERSION!!]
|
||||
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
% PLEASE refer to the accompanying document:
|
||||
% "READ_ME_App_1B_v1_Description_Cluster_Analysis.docx"
|
||||
% for description of the Application and its requirements.
|
||||
% --------------------------------------------------------------------------------------------------------
|
||||
% DSCRIPTION: The function is actually a compilation of existing and well-
|
||||
% known clustering algorithms available within the MATLAB
|
||||
% libraries, therefore the corresponding functions,descriptions
|
||||
% information and references can be retrieved from the Matlab
|
||||
% help. The function used are "kmeans", "linkage", "cluster" &
|
||||
% "fcm".
|
||||
% NOTE: Working in the Equivalent Dimension phase space,
|
||||
% leads to the usage of Euclidean distance metric, therefore all
|
||||
% other available metrics are disregarded within this Function.
|
||||
% --------------------------------------------------------------------------------------------------------
|
||||
% INPUT: The function takes as input the output file generated after running
|
||||
% <EFBFBD>ED_ToolBox_Wrapper.mat", therefore all analyses are performed
|
||||
% within the Equivalent Dimension phase space:
|
||||
% --- Tdata: is the output of "ED_ToolBox_Wrapper.mat" function,
|
||||
% corresponding to the dataset with parameters (Seismic/
|
||||
% Production) after they are transformed to ED space.
|
||||
% --- vectors: The User is requested to specify the columns from "Tdata"
|
||||
% structure, to be used in the Cluster Analysis
|
||||
% - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|
||||
% The User is then requested to enter values for some additional parameters.
|
||||
% Input Parameters Overview:
|
||||
% --- N: number of clusters to be constructed after the analysis
|
||||
% (default: 4, however, this is a completely arbitrary selection)
|
||||
% --- Meth: Clustering Algorithm Selection.
|
||||
% Possible arguments: 'Partitioning','Hierarchical' and 'Fuzzy'
|
||||
% --- CTMeth: Cluster Tree method
|
||||
% Possible arguments:'average<EFBFBD>,<EFBFBD>centroid<EFBFBD>,<EFBFBD>complete<EFBFBD>,<EFBFBD>median<EFBFBD>,
|
||||
% <EFBFBD>single<EFBFBD>,<EFBFBD>ward<EFBFBD>, and <EFBFBD>weighted<EFBFBD>
|
||||
% NOTE: CTMeth is only applicable for Meth='Hierarchical'
|
||||
% --- Lnodes: number of leaf nodes for plotting the Ward Diagram
|
||||
% (This applies only in the visualization option)
|
||||
% ---------------------------------------------------------------------------------------------------------
|
||||
% OUTPUTS: <> Cluster: Structure, with a size corresponding to the number
|
||||
% of clusters set by the use. It consists of the 3 following fields
|
||||
% - Cluster.id --> field with the parameters for each cluster
|
||||
% - Cluster.index --> field with index of the events comprising the
|
||||
% clusters for reference to the input data.
|
||||
% - Cluster.Center --> Center of mass of each Cluster
|
||||
% --------------------------------------------------------------------------------------------------------
|
||||
% <> ClusterColumns --> string array with transformed parameters
|
||||
% corresponing to the columns of Cluster.id
|
||||
% <> *** L --> Array (double) used for visualization of the Dendrogram
|
||||
% (valid only when Meth='Hierarchical'
|
||||
% --------------------------------------------------------------------------------------------------------
|
||||
% <> FIGURES:
|
||||
% - Cluster_output.jpg: Only valid for 2D and 3D cases
|
||||
% - Dendrogram_uutput.jpg: Ward diagram, only valid when Meth='Hierarchical' is selected
|
||||
% --------------------------------------------------------------------------------------------------------
|
||||
% <> Output ASCII Files: For convenience of the User, three additional ASCII
|
||||
% files are produced with the main results of the Application:
|
||||
% - Clusters_Original.txt : File containing the Original parameters of Seismic and
|
||||
% Production data selected for the Analysis. Each
|
||||
% column corresponds to the values of one parameter,
|
||||
% which is defined in <EFBFBD>Data_Fields.txt<EFBFBD> output file. THE
|
||||
% LAST COLUMN in the file, is an integer corresponding
|
||||
% to the number of cluster that the eventbelongs to.
|
||||
% NOTE!!!! that the file contains FULL (original) SAMPLE,
|
||||
% therefore NANS may be included.
|
||||
% - Clusters_Transformed.txt : File containing Transformed parameters of Seismic
|
||||
% and Production data derived by the Analysis. Each
|
||||
% column corresponds to the values of one parameter,
|
||||
% which is defined in <EFBFBD>Data_Fields.txt<EFBFBD> output file.
|
||||
% NOTE!!! that the file contains only the values that
|
||||
% were transformed (non-NANs)
|
||||
% - Clusters_Fields.txt : File containing the labels (Fileds) of each column of the
|
||||
% previously described output (ASCII) files.
|
||||
% ---------------------------------------------------------------------------------------------------------
|
||||
% LICENSE
|
||||
% This is free software: you can redistribute it and/or modify it under
|
||||
% the terms of the GNU General Public License as published by the
|
||||
% Free Software Foundation, either version 3 of the License, or
|
||||
% (at your option) any later version.
|
||||
%
|
||||
% This program is distributed in the hope that it will be useful, but
|
||||
% WITHOUT ANY WARRANTY; without even the implied warranty
|
||||
% of MERCHANTABILITY or FITNESS FOR A PARTICULAR
|
||||
% PURPOSE. See the GNU General Public License for more details.
|
||||
% ---------------------------------------------------------------------------------------------------------
|
||||
|
||||
clear; clc;
|
||||
close all
|
||||
|
||||
mkdir Outputs_Clustering
|
||||
|
||||
% ------- Path to the data directory -------
|
||||
cd Transformed_Data
|
||||
d=dir;
|
||||
dstr={d.name};
|
||||
% -----------------------------------------------
|
||||
|
||||
% -------- Select Input Data File & ---------
|
||||
% ----- Parameters to be clustered -------
|
||||
[s,s1]=DatLoad(dstr);
|
||||
load(dstr{s});
|
||||
% ----------------------------------------------
|
||||
cd ../
|
||||
% FIRST OF ALL, lets have the events index related to the input catalog!!!!
|
||||
% ...
|
||||
|
||||
% Show in Screen Clustering Options
|
||||
%open('Cluster_tree.pdf')
|
||||
% A=imread('Cluster_tree.jpg');
|
||||
% imshow(A)
|
||||
% ----------------------------------------
|
||||
|
||||
% prepare data and eliminate Nans'
|
||||
a1=[];
|
||||
for i=1:length(s1)
|
||||
a1=[a1,Tdata(s1(i)).all];
|
||||
ClusterColumns{i}=Tdata(s1(i)).field;
|
||||
end
|
||||
|
||||
%[indall,NP]=size(a1);
|
||||
a=[];cou=0;indx=zeros(1000000,1);
|
||||
for i=1:length(a1);
|
||||
if isnan(prod(a1(i,:)))==0;a=[a;a1(i,:)];cou=cou+1;indx(cou)=i;end
|
||||
end
|
||||
indx=indx';
|
||||
|
||||
% SET THE NUMBER OF CLUSTERS
|
||||
N=input('Please set the number of clusters: ');
|
||||
if cou<=N;error(['The number of events (',num2str(cou),') is smaller than the selected number of clusters (',num2str(N),')']);end
|
||||
|
||||
% ------- Select Clustering Method --------
|
||||
% Set clustering Method, among:
|
||||
% 'Partitioning' - 1, 'Hierarchical' -2, 'Fuzzy' -3
|
||||
|
||||
[Meth,ok]=listdlg('PromptString','Select Clustering Algorithm:',...
|
||||
'ListString',{'Partitioning','Hierarchical','Fuzzy'},'ListSize',[160,100],'SelectionMode','single');
|
||||
|
||||
switch Meth
|
||||
|
||||
case 1 % for 'Partitioning'
|
||||
% the following lines are disabled because Euclidean Metric applies in ED phase space
|
||||
% Set DIS, i.e. Distance Metric among:
|
||||
%metrics={'sqeuclidean','cityblock','cosine'};
|
||||
%[DIS,ok]=listdlg('PromptString','Select Distance Metric:',...
|
||||
% 'ListString',metrics,'ListSize',[160,100],'SelectionMode','single');
|
||||
|
||||
[c,centloc,sd]=kmeans(a,N,'Distance','sqeuclidean');
|
||||
|
||||
|
||||
case 2 % for 'Hierarchical'
|
||||
% Set method for computation of cluster tree among:
|
||||
|
||||
CTCmeth={'average','centroid','complete','median','single','ward','weighted'};
|
||||
[CMeth,ok]=listdlg('PromptString','Select Method for Cluster Tree Computation:',...
|
||||
'ListString',CTCmeth,'ListSize',[160,100],'SelectionMode','single');
|
||||
|
||||
% the following lines are disabled because Euclidean Metric applies in ED phase space
|
||||
% Set the Distance metric among [select only the most important ones]
|
||||
%metrics={'euclidean','squaredeuclidean','seuclidean','mahalanobis','minkowski',...
|
||||
% 'chebychev','cosine','correlation','spearman'};
|
||||
%[DIS,ok]=listdlg('PromptString','Select Distance Metric:',...
|
||||
% 'ListString',metrics,'ListSize',[160,100],'SelectionMode','single');
|
||||
|
||||
L=linkage(a,CTCmeth(CMeth),'euclidean');
|
||||
c=cluster(L,N);
|
||||
%c=clusterdata(a,'maxclust',N);
|
||||
% Consider Plotting the Dendrogram!!
|
||||
|
||||
case 3 %for 'Fuzzy'
|
||||
% Needs to remove Nans!!
|
||||
|
||||
[cen,u] = fcm(a,N);
|
||||
maxu=max(u);
|
||||
for i=1:N
|
||||
ind=find(u(i,:)==maxu);
|
||||
c(ind)=i;
|
||||
end
|
||||
c=c';
|
||||
|
||||
end
|
||||
|
||||
|
||||
|
||||
% % EXTRACT CLUSTERS
|
||||
|
||||
for j=1:N
|
||||
Cluster(j).id=a(c==j,:);
|
||||
Cluster(j).index=indx(c==j);
|
||||
Cluster(j).Center=mean(Cluster(j).id);
|
||||
if numel(Cluster(j).Center)~=numel(s1);Cluster(j).Center=Cluster(j).id;end
|
||||
end
|
||||
|
||||
%% Create Output ASCII FILE:
|
||||
a(:,size(a,2)+1)=c;
|
||||
for j=1:length(s1);b(:,j)=Tdata(s1(j)).origval_all;end
|
||||
nb=size(b,2)+1;
|
||||
for j=1:size(Cluster,2);b(Cluster(j).index,nb)=j;end
|
||||
|
||||
|
||||
|
||||
%% Store Outputs
|
||||
cd Outputs_Clustering\
|
||||
save('Cluster.mat','Cluster')
|
||||
save('ClusterColumns.mat','ClusterColumns')
|
||||
fid1=fopen('Clusters_Transformed.txt','w');
|
||||
fid2=fopen('Clusters_Original.txt','w');
|
||||
fid3=fopen('Clusters_Fields.txt','w');
|
||||
fprintf(fid1,[repmat('%32.16f ',[1,size(a,2)-1]),' %d \n'],a');
|
||||
fprintf(fid2,[repmat('%32.16f ',[1,size(b,2)-1]),' %d \n'],b');
|
||||
for i=1:size(ClusterColumns,2)
|
||||
fprintf(fid3,'%s ',ClusterColumns{i});
|
||||
end
|
||||
|
||||
fclose all;
|
||||
cd ../
|
||||
|
||||
%% Plotting example
|
||||
if length(s1)==2
|
||||
for j=1:N
|
||||
plot(Cluster(j).id(:,1),Cluster(j).id(:,2),'o');hold on;axis square
|
||||
plot(Cluster(j).Center(1),Cluster(j).Center(2),'kx','MarkerSize',16,'LineWidth',2)
|
||||
end
|
||||
xlabel({Tdata(s1(1)).field},'FontSize',14,'Interpreter','none')
|
||||
ylabel({Tdata(s1(2)).field},'FontSize',14,'Interpreter','none')
|
||||
cd Outputs_Clustering\;saveas(gcf,'Cluster_output.jpg');cd ../
|
||||
|
||||
elseif length(s1)==3
|
||||
for j=1:N
|
||||
plot3(Cluster(j).id(:,1),Cluster(j).id(:,2),Cluster(j).id(:,3),'o');hold on;grid on;axis square
|
||||
plot3(Cluster(j).Center(1),Cluster(j).Center(2),Cluster(j).Center(3),'kx','MarkerSize',16,'LineWidth',2)
|
||||
end
|
||||
xlabel({Tdata(s1(1)).field},'FontSize',14,'Interpreter','none')
|
||||
ylabel({Tdata(s1(2)).field},'FontSize',14,'Interpreter','none')
|
||||
zlabel({Tdata(s1(3)).field},'FontSize',14,'Interpreter','none')
|
||||
cd Outputs_Clustering\;saveas(gcf,'Cluster_output.jpg');cd ../
|
||||
|
||||
end
|
||||
|
||||
|
||||
if Meth==2;
|
||||
Lnodes=dialog1('number of leaf nodes',{num2str(size(c,1))});
|
||||
figure;dendrogram(L,Lnodes,'ColorThreshold', L(length(L)-N+2,3));
|
||||
cd Outputs_Clustering\;saveas(gcf,'Dendrogram_output.jpg');cd ../;end
|
||||
|
||||
% *************************** FUNCTIONS ***************************
|
||||
% ****-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-****
|
||||
%% --------------------------------------------------------------------------------------
|
||||
%Select Seismic Catalog and the parameters the user wishes to study
|
||||
function [s,s1]=DatLoad(dstr)
|
||||
% Select Seismic Catalog
|
||||
[s,ok]=listdlg('PromptString','Select Transformed Dataset:',...
|
||||
'SelectionMode','single',...
|
||||
'ListString',dstr);
|
||||
if ok; load(dstr{s});end
|
||||
|
||||
%Select Parameters from Seismic Catalog
|
||||
[s1,ok]=listdlg('PromptString','Select field(s):',...
|
||||
'ListString',{Tdata.field});
|
||||
|
||||
end
|
||||
%%
|
||||
function [ou]=dialog1(name,defaultanswer)
|
||||
|
||||
prompt=['\fontsize{12} Please enter ',name, ':'];
|
||||
prompt={prompt};
|
||||
numlines=1; opts.Interpreter='tex';
|
||||
ou=inputdlg(prompt,name,numlines,defaultanswer,opts);ou=str2num(ou{1});
|
||||
|
||||
end
|
||||
|
||||
Binary file not shown.
Binary file not shown.
Binary file not shown.
Reference in New Issue
Block a user