Putri, Marina Marsudi (2015) Pengelompokan Kabupaten/Kota Di Jawa Timur Berdasarkan Indikator Kesehatan Masyarakat Menggunakan Metode Kohonen Som Dan K-Means. Undergraduate thesis, Institut Technology Sepuluh Nopember.
Preview |
Text
1311100028-Undergraduate Theses.pdf - Published Version Download (1MB) | Preview |
Abstract
Kondisi kesehatan masyarakat di Jawa Timur yang tidak homogen akan menyulitkan
bagi Tim Pembina Kota Sehat pada saat melakukan pembinaan dan monitoring.
Berdasarkan hal tersebut diperlukan pengelompokan kabupaten dan kota di
Jawa Timur yang didasarkan pada kemiripan karakteristik kondisi kesehatan masyarakat,
sehingga proses pembinaan dan monitoring akan dilakukan berdasarkan
kelompok kabupaten dan kota yang terbentuk. Pengelompokan kabupaten/kota di
Jawa Timur akan dilakukan menggunakan metode Kohonen SOM kemudian akan
dibandingkan dengan hasil pengelompokan metode klasik K-Means, yang merupakan
metode pengelompokan yang populer dan sering digunakan, berdasarkan
kriteria nilai icdrate (internal cluster dispersion rate). Pada pengujian Bartlett
diperoleh kesimpulan bahwa terdapat hubungan atau korelasi di antara variabel
penelitian. Setelah dilakukan analisis faktor untuk mereduksi variabel diperoleh 3
faktor baru yang terbentuk. Berdasarkan nilai Pseudo Fstatistics yaitu sebesar
13,819, hasil pengelompokan terbaik adalah menggunakan metode Kohonen SOM
dengan jenis topologi hextop. Sedangkan pada metode K-Means nilai Pseudo Fstatistics
terbesar yaitu 9,781 ketika digunakan kelompok sebanyak 2 klaster. Perbandingan
hasil klaster terbaik berdasarkan nilai icdrate diperoleh kesimpulan bahwa
nilai icdrate metode Kohonen SOM yaitu sebesar 0,962 lebih kecil dibandingkan
dengan nilai icdrate hasil pengelompokan metode K-Means yaitu sebesar 0,988. Pada
hasil pengujian One-way MANOVA diperoleh kesimpulan bahwa pada masingmasing
kelompok yang terbentuk memiliki perbedaan.
====================================================================================================
Public health conditions in East Java is not homogeneous. That will be
difficult for Team evaluation to monitoring about public health condition in
that area. Based on the required, grouping of districts and cities in East
Java, which is based on the similarity of the characteristics of public health
conditions, so that the process of monitoring will be done by district and
city groups are formed. The grouping of districts/cities in East Java will
done by using Kohonen SOM and then will compared with the results of
grouping classic K-Means method, based on the criteria of value icdrate
(internal cluster dispersion rate). In the Bartlett test is concluded that there
is a relationship or correlation between the study variables. After analyzing
the factors to reduce the variables obtained three new factors were formed.
Based on the value of the Pseudo Fstatistics that is 13,819, the best result
for clustering using Kohonen SOM with hextop topolgy. Beside that the
value of Pseudo Fstatistics using K-Means with 2 cluster is 9,871. The
result of the comparison from clustering based on icdrate value is Kohonen
SOM has value 0,962 is smaller than icdrate value from K-Means that is
0,988. From One-way NAMOVA testing the conclusion is cluster that
performed has different characteristics.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSSt 519.53 Put p |
| Uncontrolled Keywords: | Icdrate, Indikator Kesehatan Masyarakat, K-means, Kohonen SOM, One-way MANOVA, Pseudo F-statistics |
| Subjects: | Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis. Correspondence analysis (Statistics) |
| Divisions: | Faculty of Mathematics and Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Mr. Tondo Indra Nyata |
| Date Deposited: | 24 May 2018 04:04 |
| Last Modified: | 24 May 2018 04:04 |
| URI: | http://repository.its.ac.id/id/eprint/51943 |
Actions (login required)
 |
View Item |