A- 1 LAMPIRAN A KLUSTER SOM DAN VALIDASI RMSSTD A. Algortma SOM Berkut n merupakan source code algortma SOM kluster 3 kluster 6: B. Perhtungan Kluster Berkut n merupakan source code untuk kluster B, yang haslnya memlk 3 kluster Source Code Kluster 3 SOM % nsalsas data ht=0.6; % melakukan random weght weght1=[0.1875 0.7690 0.3960 0.4600]; weght2=[0.2729 0.4517 0.6099 0.7421]; weght3=[0.0594 0.1253 0.1302 0.4574]; for =1:4000; a=(weght1(1,1)-data(,1))^2+(weght1(1,2)- data(,2))^2+(weght1(1,3)-data(,3))^2; b=(weght2(1,1)-data(,1))^2+(weght2(1,2)- data(,2))^2+(weght2(1,3)-data(,3))^2; c=(weght3(1,1)-data(,1))^2+(weght3(1,2)- data(,2))^2+(weght3(1,3)-data(,3))^2; f a<b && a<c z=data(,:)-weght1(1,:); weght1=weght1(1,:)+(ht*z); weght2=weght2(1,:); weght3=weght3(1,:);
else f b<c && b<a z=data(,:)-weght2(1,:); weght2=weght2(1,:)+(ht*z); weght1=weght1(1,:); weght3=weght3(1,:); else z=data(,:)-weght3(1,:); weght3=weght3(1,:)+(ht*z); weght1=weght1(1,:); weght2=weght2(1,:); dsp('update weght berhasl') 2 weght1 weght2 weght3 for j=1:4000; a=(weght1(1,1)-data(j,1))^2+(weght1(1,2)- data(j,2))^2+(weght1(1,3)-data(j,3))^2; b=(weght2(1,1)-data(j,1))^2+(weght2(1,2)- data(j,2))^2+(weght2(1,3)-data(j,3))^2; c=(weght3(1,1)-data(j,1))^2+(weght3(1,2)- data(j,2))^2+(weght3(1,3)-data(j,3))^2; f a<b && a<c dsp ('1') else f b<c && b<a dsp ('2') C. else Kluster C dsp ('3')
A- 3 Berkut n merupakan source code kluster c, yang memlk hasl 4 kluster Source Code Kluster 4 % nsalsas Data ht=0.6; %melakukan random weght weght1=[0.1875 0.7690 0.3960 0.4600]; weght2=[0.2729 0.4517 0.6099 0.7421]; weght3=[0.0594 0.1253 0.1302 0.4574]; weght4=[0.0924 0.7229 0.5312 0.4327]; for =1:4000; a=(weght1(1,1)-data(,1))^2+(weght1(1,2)- Data(,2))^2+(weght1(1,3)-Data(,3))^2+(weght1(1,4)- Data(,4))^2; b=(weght2(1,1)-data(,1))^2+(weght2(1,2)- Data(,2))^2+(weght2(1,3)-Data(,3))^2+(weght1(1,4)- Data(,4))^2; c=(weght3(1,1)-data(,1))^2+(weght3(1,2)- Data(,2))^2+(weght3(1,3)-Data(,3))^2+(weght1(1,4)- Data(,4))^2; d=(weght4(1,1)-data(,1))^2+(weght4(1,2)- Data(,2))^2+(weght4(1,3)-Data(,3))^2+(weght1(1,4)- Data(,4))^2;
4 else f b<c && b<d && b<a z=data(,:)-weght2(1,:); weght2=weght2(1,:)+(ht*z); weght1=weght1(1,:); weght3=weght3(1,:); weght4=weght4(1,:); else f c<d && c<a && c<b z=data(,:)-weght3(1,:); weght3=weght3(1,:)+(ht*z); weght1=weght1(1,:); weght2=weght2(1,:); weght4=weght4(1,:); else z=data(,:)-weght4(1,:); weght4=weght4(1,:)+(ht*z); weght1=weght1(1,:); weght2=weght2(1,:); weght3=weght3(1,:); dsp ('update weght berhasl')
A- 5 weght1 weght2 weght3 weght4 for j=1:4000; a=(weght1(1,1)-data(j,1))^2+(weght1(1,2)- Data(j,2))^2+(weght1(1,3)-Data(j,3))^2+(weght1(1,4)- Data(j,4))^2; b=(weght2(1,1)-data(j,1))^2+(weght2(1,2)- Data(j,2))^2+(weght2(1,3)-Data(j,3))^2+(weght1(1,4)- Data(j,4))^2; c=(weght3(1,1)-data(j,1))^2+(weght3(1,2)- Data(j,2))^2+(weght3(1,3)-Data(j,3))^2+(weght1(1,4)- Data(j,4))^2; d=(weght4(1,1)-data(j,1))^2+(weght4(1,2)- Data(j,2))^2+(weght4(1,3)-Data(j,3))^2+(weght1(1,4)- Data(j,4))^2; f a<b && a<c && a<d dsp ('1') else f b<c && b<d && b<a dsp ('2') else f c<d && c<a && c<b dsp ('3') else dsp ('4')
6 D. Kluster D Berkut n merupakan source code kluster d, yang memlk hasl 5 kluster Source Code Kluster d data ht=0.6; %melakukan random weght weght1=[0.1875 0.7690 0.3960 0.4600]; weght2=[0.2729 0.4517 0.6099 0.7421]; weght3=[0.0594 0.1253 0.1302 0.4574]; weght4=[0.0924 0.7229 0.5312 0.8327]; weght5=[0.1088 0.0986 0.1420 0.7521]; for =1:4000; a=(weght1(1,1)-data(,1))^2+(weght1(1,2)- data(,2))^2+(weght1(1,2)-data(,2))^2+(weght1(1,2)- data(,2))^2; b=(weght2(1,1)-data(,1))^2+(weght2(1,2)- data(,2))^2+(weght1(1,2)-data(,2))^2+(weght1(1,2)- data(,2))^2; c=(weght3(1,1)-data(,1))^2+(weght3(1,2)- data(,2))^2+(weght1(1,2)-data(,2))^2+(weght1(1,2)- data(,2))^2; d=(weght4(1,1)-data(,1))^2+(weght4(1,2)- data(,2))^2+(weght1(1,2)-data(,2))^2+(weght1(1,2)- data(,2))^2; e=(weght5(1,1)-data(,1))^2+(weght5(1,2)- data(,2))^2+(weght1(1,2)-data(,2))^2+(weght1(1,2)- data(,2))^2; data_temp = [a b c d e];
f mn(data_temp)==a; z=data(,:)-weght1(1,:); weght1=weght1(1,:)+(ht*z); weght2=weght2(1,:); weght3=weght3(1,:); weght4=weght4(1,:); weght5=weght5(1,:); else f mn(data_temp)==b; z=data(,:)-weght2(1,:); weght2=weght2(1,:)+(ht*z); weght1=weght1(1,:); weght3=weght3(1,:); weght4=weght4(1,:); weght5=weght5(1,:); else f mn(data_temp)==c; z=data(,:)-weght3(1,:); weght3=weght3(1,:)+(ht*z); weght1=weght1(1,:); weght2=weght2(1,:); weght4=weght4(1,:); weght5=weght5(1,:); else f mn(data_temp)==d; z=data(,:)-weght4(1,:); weght4=weght4(1,:)+(ht*z); weght5=weght5(1,:); weght1=weght1(1,:); weght2=weght2(1,:); weght3=weght3(1,:); else f mn(data_temp)==e; z=data(,:)-weght5(1,:); weght5=weght5(1,:)+(ht*z); weght4=weght4(1,:); weght1=weght1(1,:); weght2=weght2(1,:); weght3=weght3(1,:); A- 7
8 end end end end end end dsp('update weght berhasl') weght1 weght2 weght3 weght4 weght5 for j=1:4000; a=(weght1(1,1)-data(j,1))^2+(weght1(1,2)- data(j,2))^2+(weght1(1,3)-data(j,3))^2+(weght1(1,4)- data(j,4))^2; b=(weght2(1,1)-data(j,1))^2+(weght2(1,2)- data(j,2))^2+(weght1(1,3)-data(j,3))^2+(weght1(1,4)- data(j,4))^2; c=(weght3(1,1)-data(j,1))^2+(weght3(1,2)- data(j,2))^2+(weght1(1,3)-data(j,3))^2+(weght1(1,4)- data(j,4))^2; d=(weght4(1,1)-data(j,1))^2+(weght4(1,2)- data(j,2))^2+(weght1(1,3)-data(j,3))^2+(weght1(1,4)- data(j,4))^2; e=(weght5(1,1)-data(j,1))^2+(weght5(1,2)- data(j,2))^2+(weght1(1,3)-data(j,3))^2+(weght1(1,4)- data(j,4))^2;
A- 9 f (a<b && a<c && a<d && a<e) dsp ('1') else f (b<c && b<d && b<e && b<a) dsp ('2') else f (c<d && c<e && c<a && c<b) dsp ('3') else f (d<e && d<a && d<b && d<c) dsp ('4') else dsp ('5')
10 E. Kluster E Source Code Kluster 6 % nsalsas data ht=0.6; % melakukan random weght weght1=[0.1875 0.7690 0.3960 0.4600]; weght2=[0.2729 0.4517 0.6099 0.7421]; weght3=[0.0594 0.1253 0.1302 0.4574]; weght4=[0.0924 0.7229 0.5312 0.8327]; weght5=[0.1088 0.0986 0.1420 0.7521]; weght6=[0.1683 0.2176 0.2510 0.8472]; for =1:4000; a=(weght1(1,1)-data(,1))^2+(weght1(1,2)- data(,2))^2+(weght1(1,3)-data(,3))^2+(weght1(1,4)- data(,4))^2; b=(weght2(1,1)-data(,1))^2+(weght2(1,2)- data(,2))^2+(weght2(1,3)-data(,3))^2+(weght1(1,4)- data(,4))^2; c=(weght3(1,1)-data(,1))^2+(weght3(1,2)- data(,2))^2+(weght3(1,3)-data(,3))^2+(weght1(1,4)- data(,4))^2; d=(weght4(1,1)-data(,1))^2+(weght4(1,2)- data(,2))^2+(weght4(1,3)-data(,3))^2+(weght1(1,4)- data(,4))^2; e=(weght5(1,1)-data(,1))^2+(weght5(1,2)- data(,2))^2+(weght5(1,3)-data(,3))^2+(weght1(1,4)- data(,4))^2; f=(weght6(1,1)-data(,1))^2+(weght6(1,2)- data(,2))^2+(weght6(1,3)-data(,3))^2+(weght1(1,4)- data(,4))^2; data_temp = [a b c d e f];
f mn(data_temp)==a; z=data(,:)-weght1(1,:); weght1=weght1(1,:)+(ht*z); weght2=weght2(1,:); weght3=weght3(1,:); weght4=weght4(1,:); weght5=weght5(1,:); weght6=weght6(1,:); else f mn(data_temp)==b; z=data(,:)-weght2(1,:); weght2=weght2(1,:)+(ht*z); weght1=weght1(1,:); weght3=weght3(1,:); weght4=weght4(1,:); weght5=weght5(1,:); weght6=weght6(1,:); else f mn(data_temp)==c; z=data(,:)-weght3(1,:); weght3=weght3(1,:)+(ht*z); weght1=weght1(1,:); weght2=weght2(1,:); weght4=weght4(1,:); weght5=weght5(1,:); weght6=weght6(1,:); else f mn(data_temp)==d; z=data(,:)-weght4(1,:); weght4=weght4(1,:)+(ht*z); weght5=weght5(1,:); weght6=weght6(1,:); weght1=weght1(1,:); weght2=weght2(1,:); weght3=weght3(1,:); A- 11
12 else f mn(data_temp)==e; z=data(,:)-weght5(1,:); weght5=weght5(1,:)+(ht*z); weght6=weght6(1,:); weght4=weght4(1,:); weght1=weght1(1,:); weght2=weght2(1,:); weght3=weght3(1,:); else f mn(data_temp)==f; z=data(,:)-weght6(1,:); weght6=weght6(1,:)+(ht*z); weght5=weght5(1,:); weght4=weght4(1,:); weght1=weght1(1,:); weght2=weght2(1,:); weght3=weght3(1,:); dsp('update weght berhasl')
A- 13 weght1 weght2 weght3 weght4 weght5 weght6 for j=1:4000; a=(weght1(1,1)-data(j,1))^2+(weght1(1,2)- data(j,2))^2+(weght1(1,3)-data(j,3))^2+(weght1(1,4)-data(j,4))^2; b=(weght2(1,1)-data(j,1))^2+(weght2(1,2)- data(j,2))^2+(weght2(1,3)-data(j,3))^2+(weght2(1,4)-data(j,4))^2; c=(weght3(1,1)-data(j,1))^2+(weght3(1,2)- data(j,2))^2+(weght3(1,3)-data(j,3))^2+(weght3(1,4)-data(j,4))^2; d=(weght4(1,1)-data(j,1))^2+(weght4(1,2)- data(j,2))^2+(weght4(1,3)-data(j,3))^2+(weght4(1,4)-data(j,4))^2; e=(weght5(1,1)-data(j,1))^2+(weght5(1,2)- data(j,2))^2+(weght5(1,3)-data(j,3))^2+(weght5(1,4)-data(j,4))^2; f=(weght6(1,1)-data(j,1))^2+(weght6(1,2)- data(j,2))^2+(weght6(1,3)-data(j,3))^2+(weght6(1,4)-data(j,4))^2; f a<b && a<c && a<d && a<e && a<f dsp ('1') else f b<c && b<d && b<e && b<f && b<a dsp ('2') else f c<d && c<e && c<f && c<a && c<b dsp ('3') else f d<e && d<f && d<a && d<b && d<c dsp ('4') else f e<f && e<a && e<b && e<c && e<d dsp ('5') else dsp ('6')
14 Hasl Algortma SOM : Jumlah Kluster 2 Jumlah Kluster 3 Jumlah Kluster 4 Jumlah Kluster 5 Jumlah Kluster 6 Kluster 2 Kluster 3 Kluster 4 Kluster 5 Kluster 6 1823 1801 1801 0 0 2177 1699 1699 1699 1699 500 500 1801 1801 0 0 0 500 500 F. Valdas RMSSTD Perhtungan valdas RMSSTD: Tabel Perhtungan RMSSTD Kluster A Rata2 (Kluster 1) Rata2 (kluster 2) jumlah(x-x) pd kluster 1 jumlah(x-x) pd kluster 2 1.0272254 0.377964 1.026148 0.812861-0.860188-0.3165-0.85929-0.68068 2662.426 3584.401
A- 15 nj pada 1823 1823 1823 1823 7292 kluster 1 dmens... nj pada 2177 2177 2177 2177 8708 kluster 2 dmens... RMSSTD 0.6248804 Tabel Perhtungan RMSSTD Kluster B Rata2 (kluster 1) -0.82968 0.377964-0.85929-0.59349 Rata2 (kluster 2) 1.16275 0.377964 1.163755 0.881045 Rata2 (kluster 3) -0.96251-2.64575-0.85929-0.85606 jumlah(x-x) pd 65.90982833 kluster 1 jumlah(x-x) pd 1614.10295 kluster 2 jumlah(x-x) pd 0.379604129 kluster 3 nj pada kluster 1 1801 1801 1801 1801 7204 dmens... nj pada kluster 2 1699 1699 1699 1699 6796 dmens... nj pada kluster 3 500 500 500 500 2000 dmens... RMSSTD 0.3241053
16 Tabel Perhtungan RMSSTD Kluster C Rata2 (kluster1 ) -0.82968 0.377964-0.85929-0.59349 Rata2 (kluster 2) 1.16275 0.377964 1.163755 0.881045 Rata2 (kluster 3) -0.96251-2.64575-0.85929-0.85606 Rata2 (kluster 4) 0 0 0 0 jumlah(x-x) pd kluster 65.90982833 1 jumlah(x-x) pd kluster 1614.10295 2 jumlah(x-x) pd kluster 0.379604129 3 jumlah(x-x) pd kluster 4 nj pada kluster 1 1801 1801 1801 1801 7204 dmens... nj pada kluster 2 1699 1699 1699 1699 6796 dmens... nj pada kluster 3 500 500 500 500 2000 dmens... nj pada kluster 4 0 0 0 0 0 dmens... RMSSTD 0.3241154
A- 17 Tabel Perhtungan RMSSTD Kluster D Rata2 (kluster 1) 0 0 0 0 Rata2 (kluster 2) 1.16275 0.377964 1.163755 0.88104541 Rata2 (kluster 3) -0.82968 0.377964-0.85929-0.5934863 Rata2 (kluster 4) 0 0 0 Rata2 (kluster 5) -0.96251-2.64575-0.85929-0.8560592 jumlah(x-x) pd kluster 1 0 jumlah(x-x) pd kluster 2 1614.10295 jumlah(x-x) pd kluster 3 65.90982833 jumlah(x-x) pd kluster 4 0 jumlah(x-x) pd kluster 5 0.379604129 nj pada kluster 1 dmens... 0 0 0 0 nj pada kluster 2 dmens... 1699 1699 1699 1699 6796 nj pada kluster 3 dmens... 1801 1801 1801 1801 7204 nj pada kluster 4 dmens... 0 0 0 0 0 nj pada kluster 5 dmens... 500 500 500 500 2000 RMSSTD 0.324125523
18 Tabel Perhtungan RMSSTD Kluster E Rata2 (kluster 1) 0 0 0 0 Rata2 (kluster 2) 1.16275 0.377964 1.163755 0.881045 Rata2 (kluster 3) -0.82968 0.377964-0.85929-0.59349 Rata2 (kluster 4) 0 0 0 Rata2 (kluster 5) -0.96251-2.64575-0.85929-0.85606 Rata2 (kluster 6) 0 0 0 0 jumlah(x-x) pd kluster 1 0 jumlah(x-x) pd kluster 2 1614.035 jumlah(x-x) pd kluster 3 65.89382 jumlah(x-x) pd kluster 4 0 jumlah(x-x) pd kluster 5 0.378869 jumlah(x-x) pd kluster 6 0 nj pada kluster 1 dmens... 0 0 0 0 nj pada kluster 2 dmens... 1699 1699 1699 1699 6796 nj pada kluster 3 dmens... 1801 1801 1801 1801 7204 nj pada kluster 4 dmens... 0 0 0 0 0 nj pada kluster 5 dmens... 499 499 499 499 1996 nj pada kluster 6 dmens... 0 0 0 0 0 RMSSTD 0.324168
A-19 Hasl Valdas RMSSTD A- 1 Tabel Hasl RMSSTD RMSSTD Nla Kluster a 0.6248804 Kluster b 0.3241053 Kluster c 0.3241154 Kluster d 0.3241255 Kluster e 0.324168 Grafk Hasl RMSSTD A- 2 Grafk Hasl RMSSTD
B-1 20 LAMPIRAN B Kluster Kmeans beserta Valdas DBI A. Source Code Algortma Kombnas SOM dan Kmeans Ketk pada Matlab [klaster, ctrs]=kmeans(data,3) Segmen Kode K-Means Hasl Dar Algortma Kmeans Hasl dar Algortma K-means
A-21 B-2 Hasl Algortma Kmeans Tabel Perhtungan algortma K-means Hasl Kluster 1 Hasl Kluster 2 Hasl Kluster 3 Kluster 1 Kluster 2 Kluster 3 500 1801 1699 B. Hasl Valdas Daves Bouldn Index Tabel Hasl Valdas DBI Klaster Nla DBI SOM 0.26902 SOM + Kmeans 0.184783