Tampilkan postingan dengan label Statistik. Tampilkan semua postingan
Tampilkan postingan dengan label Statistik. Tampilkan semua postingan
Selasa, 11 Januari 2022
Indonesian Population Growth and Internal Migration in 2020
Population Density of 2010 (below) and 2020 (above)
Internal Recent Migration Flows of 2020
Notes:
Exact location of each city may not precisely located because clustering proces shift the center of the involved area
Selasa, 19 Maret 2019
Senin, 23 Juli 2018
Dampak Post-Suburbanisasi Dan Pertumbuhan Perkotaan di Kawasan Pinggiran Metropolitan Jabodetabek Terhadap Kerentanan Bencana Banjir
Jurnal Green Growth dan Manajemen Lingkungan Vol. 7 (1), p. 1-21
Senin, 09 April 2018
Beyond the Early Stage of Post-suburbanization: Evidence from Urban Spatial Transformation in Jabodetabek Metropolitan Area
refined version of the paper present in The 4th PlanoCosmo International Conference: “Transforming beyond Borders, Starting the New Urban Agenda”, 2-4 April 2018, Bandung, Jawa Barat
Kamis, 05 April 2018
Perbandingan Pendekatan Reformasi Metropolitan Dalam Menyelesaikan Isu Komuter di Jakarta Metropolitan Area
PLANO MADANI Volume 7 Nomor 1 April 2018, p. 89-105
Jumat, 02 Desember 2016
Rabu, 19 Oktober 2016
Kamis, 03 Maret 2016
Senin, 11 Januari 2016
Rabu, 23 Desember 2015
Senin, 19 Oktober 2015
Minggu, 11 Januari 2015
Selasa, 23 Desember 2014
MARS And Truncated Spline Approach On Modelling Human Development Index (HDI) In Indonesia
submitted papper on 2014 International Conference on Statistics and Mathematics (ICSM 2014)
http://www.icsm.its.ac.id/
Uploaded with permission of Mr. Ayub Parlin Ampulembang
http://www.icsm.its.ac.id/
Uploaded with permission of Mr. Ayub Parlin Ampulembang
Senin, 01 Desember 2014
Pemanfaatan Data Susenas pada pemodelan rasio keluarga pra sejahtera
Papper pemenang beasiswa IPADI
dipresentasikan dalam Seminar Ilmiah Nasional Kependudukan 2014
Bandung 27 November 2014
http://www.seminar-kependudukan.org/
dipresentasikan dalam Seminar Ilmiah Nasional Kependudukan 2014
Bandung 27 November 2014
http://www.seminar-kependudukan.org/
Senin, 06 Oktober 2014
Senin, 15 September 2014
Senin, 03 Maret 2014
Senin, 24 Februari 2014
Senin, 23 Desember 2013
Minggu, 15 September 2013
Matlab Code of PCA-Part for K-Means initialization
function idx=PCAPartKmeans(data,k,method)
% input:
% k= number of clusters
% method:
% 1=HHQL (Householder-QL), is modified function taken from myEig in LAPACK (http://eriesadewo.blogspot.com/2014/05/matlab-code-of-householder-ql-method.html)
% 2=simplified truncated power method (http://eriesadewo.blogspot.com/2013/08/simplified-code-of-truncated-power.html),
% output:
% data indexed on cluster number
% Source:
% Su, T., Dy, J., 2004. A Deterministic Method for Initializing K-means Clustering. In: Proc. 16th IEEE International Conference on Tools with Tools with Artificial Intelligence (ICTAI2004), pp.784-786.
% Last Modified: 1 Februari 2014
% Erie Sadewo & A. Syahrul Choir
% erie@bps.go.id
% The user of this function is requested to cite:
% Sadewo, E., Mashuri, M., Barakbah, A.R., 2014. Pengaruh penggunaan metode power dan truncated power pada PCA-Part untuk inisialisasi K-Means. Prosiding Seminar Nasional Sains & Teknologi V, pp. 1326-1335.
centroid=PCAInitCentroid(data,k,method);
idx=kmeans(data,k,'Start',centroid);
function centroid=PCAInitCentroid(data,k,method)
newdata=data;
imaxSSE=1;
SSE=1;
for i=1:(k-1)
clust{imaxSSE}=[];
SSE(imaxSSE)=[];
[C1 C2 SSE1 SSE2]=pca_part(newdata,k,method);
clust{imaxSSE}=C1;
clust{i+1}=C2;
SSE(imaxSSE)=SSE1;
SSE(i+1)=SSE2;
imaxSSE=find(SSE==max(SSE));
newdata=clust{imaxSSE};
end
for i=1:k
m=size(clust{i},1);
if m==1;
centroid{i}=clust{i};
else
centroid{i}=mean(clust{i});
end
end
centroid=cell2mat((centroid)');
function [C1 C2 SSE1 SSE2]=pca_part(data,k,method)
S=cov(data);
if method==1
[V ~]=HHQL(S,k);
elseif method==2
[V ~]=tpower(S);
end
[C1 C2 SSE1 SSE2]=binary_part(data,method,V);
function [C1 C2 SSE1 SSE2]=binary_part(data,method,V)
mu=mean(data);
PC=data*V;
PCmu=mu*V;
if method==1
ind1=find(PC>PCmu);
ind2=find(PCPCmu);
end
ind3=find(PC==PCmu);
C1=data(ind1,:);
C2=data(ind2,:);
if size(C1,1)==1;
mu1=C1;
else
mu1=mean(C1);
end
if size(C2,1)==1;
mu2=C2;
else
mu2=mean(C2);
end
PCmu1=mu1*V;
PCmu2=mu2*V;
if isempty(ind3)==0;
jarak1= sqrt((PCmu1-PCmu).^2);
jarak2= sqrt((PCmu2-PCmu).^2);
if jarak1
C1=[C1;data(ind3,:)];
ind1=[ind1 ind3];
C1=data(ind1,:);
mu1=mean(C1);
else
C2=[C2;data(ind3,:)];
ind2=[ind2 ind3];
C2=data(ind2,:);
mu2=mean(C2);
end
end
SSE1= sum(sqrt(sum((C1-repmat(mu1,(size(C1,1)),1)).^2,2)).^2);
SSE2= sum(sqrt(sum((C2-repmat(mu2,(size(C2,1)),1)).^2,2)).^2);
% input:
% k= number of clusters
% method:
% 1=HHQL (Householder-QL), is modified function taken from myEig in LAPACK (http://eriesadewo.blogspot.com/2014/05/matlab-code-of-householder-ql-method.html)
% 2=simplified truncated power method (http://eriesadewo.blogspot.com/2013/08/simplified-code-of-truncated-power.html),
% output:
% data indexed on cluster number
% Source:
% Su, T., Dy, J., 2004. A Deterministic Method for Initializing K-means Clustering. In: Proc. 16th IEEE International Conference on Tools with Tools with Artificial Intelligence (ICTAI2004), pp.784-786.
% Last Modified: 1 Februari 2014
% Erie Sadewo & A. Syahrul Choir
% erie@bps.go.id
% The user of this function is requested to cite:
% Sadewo, E., Mashuri, M., Barakbah, A.R., 2014. Pengaruh penggunaan metode power dan truncated power pada PCA-Part untuk inisialisasi K-Means. Prosiding Seminar Nasional Sains & Teknologi V, pp. 1326-1335.
centroid=PCAInitCentroid(data,k,method);
idx=kmeans(data,k,'Start',centroid);
function centroid=PCAInitCentroid(data,k,method)
newdata=data;
imaxSSE=1;
SSE=1;
for i=1:(k-1)
clust{imaxSSE}=[];
SSE(imaxSSE)=[];
[C1 C2 SSE1 SSE2]=pca_part(newdata,k,method);
clust{imaxSSE}=C1;
clust{i+1}=C2;
SSE(imaxSSE)=SSE1;
SSE(i+1)=SSE2;
imaxSSE=find(SSE==max(SSE));
newdata=clust{imaxSSE};
end
for i=1:k
m=size(clust{i},1);
if m==1;
centroid{i}=clust{i};
else
centroid{i}=mean(clust{i});
end
end
centroid=cell2mat((centroid)');
function [C1 C2 SSE1 SSE2]=pca_part(data,k,method)
S=cov(data);
if method==1
[V ~]=HHQL(S,k);
elseif method==2
[V ~]=tpower(S);
end
[C1 C2 SSE1 SSE2]=binary_part(data,method,V);
function [C1 C2 SSE1 SSE2]=binary_part(data,method,V)
mu=mean(data);
PC=data*V;
PCmu=mu*V;
if method==1
ind1=find(PC>PCmu);
ind2=find(PC
end
ind3=find(PC==PCmu);
C1=data(ind1,:);
C2=data(ind2,:);
if size(C1,1)==1;
mu1=C1;
else
mu1=mean(C1);
end
if size(C2,1)==1;
mu2=C2;
else
mu2=mean(C2);
end
PCmu1=mu1*V;
PCmu2=mu2*V;
if isempty(ind3)==0;
jarak1= sqrt((PCmu1-PCmu).^2);
jarak2= sqrt((PCmu2-PCmu).^2);
if jarak1
ind1=[ind1 ind3];
C1=data(ind1,:);
mu1=mean(C1);
else
C2=[C2;data(ind3,:)];
ind2=[ind2 ind3];
C2=data(ind2,:);
mu2=mean(C2);
end
end
SSE1= sum(sqrt(sum((C1-repmat(mu1,(size(C1,1)),1)).^2,2)).^2);
SSE2= sum(sqrt(sum((C2-repmat(mu2,(size(C2,1)),1)).^2,2)).^2);
Langganan:
Postingan (Atom)


