Penggerombolan Desa Blank Spot di Kabupaten Pemalang menggunakan Metode K-Means, Fuzzy C-Means (Fcm), Pohon Keputusan dan Penskoran Terboboti

Abstract

Pandemi covid-19 berpengaruh terhadap tatanan kebiasaan masyarakat di seluruh penjuru dunia. Sektor pendidikan merupakan salah satu sektor yang terdampak dari adanya perubahan tatanan kebiasaan tersebut. Perubahan tatanan kebiasaan pada sektor pendidikan di Indonesia, tercermin dari perubahan kebijakan proses belajar mengajar secara tatap muka ke proses pembelajaran jarak jauh (PJJ / pembelajaran daring). Hal ini sejalan dengan surat edaran Menteri Pendidikan dan Kebudayaan Indonesia nomor: 36962/ MPK.A/HK/2020 tertanggal 17 Maret 2020, tentang kebijakan pembelajaran daring selama masa pandemi covid-19 untuk mengurangi penyebaran virus corona. Beberapa penelitian yang telah dilakukan untuk mengkaji kendala proses PJJ, menghasilkan temuan bahwa kualitas jaringan internet merupakan kendala mayoritas dalam proses PJJ. Kondisi dimana tidak adanya jaringan internet di suatu daerah biasa disebut dengan blank spot. Permasalahan blank spot menjadi perhatian khusus Presiden Joko Widodo. Dalam rangka meminimalisir permasalahan blank spot, Presiden dan Kominfo RI merealisasikan program “Indonesia merdeka sinyal hingga pelosok negeri”. Program ini melibatkan seluruh kabupaten/kota di Indonesia untuk melakukan survei kualitas jaringan pada area terkecil desa, serta penataan dan penyempurnaan sistem jaringan komunikasi. Kabupaten Pemalang merupakan salah satu Kabupaten di Provinsi Jawa Tengah yang melakukan survei kualitas jaringan. Survei kualitas jaringan diinisiasi oleh Pemerintah Daerah melalui Dinas Komunikasi dan Informatika serta Badan Pusat Statistik Kabupaten Pemalang, yang mencakup survei kualitas jaringan desa, jumlah penduduk, ketersediaan tower telekomunikasi, dan keadaan geografi desa. Pada dasarnya kegiatan survei kualitas jaringan memerlukan sumber daya dan biaya yang relatif tidak sedikit sehingga untuk melakukan efisiensi kegiatan survei lapangan, diperlukan pendeteksian dini status blank spot desa berdasarkan karakteristik desa blank spot secara umum. Pendekatan metode statistika yang dapat digunakan untuk menduga suatu desa termasuk dalam status blank spot atau tidak adalah dengan menggunakan metode penggerombolan. Metode penggerombolan data dapat dibedakan menjadi metode supervised classification dan metode unsupervised classification. Metode penggerombolan supervised classification yang umumnya digunakan untuk menggerombolkan desa berdasarkan karakteristik desa adalah metode pohon keputusan dan metode penskoran terboboti. Dua metode ini dipilih karena memiliki tampilan hasil gerombol yang menarik dan mudah diinterpretasi oleh pengguna dalam bentuk diagram dan skor. Sedangkan pada metode penggerombolan unsupervised classification yang umumnya digunakan untuk menggerombolkan desa berdasarkan karakteristik desa adalah metode kmeans dan metode fuzzy c-means (fcm). Dua metode ini dipilih karena memiliki kekhususan ciri untuk menggerombolkan desa berdasarkan karakteristiknya secara tegas maupun samar (memungkinkan satu desa menjadi anggota lebih dari satu gerombol). Survei kualitas jaringan di Kabupaten Pemalang telah dilakukan, namun tidak menuntut kemungkinkan akan dilakukan survei baru pada desa hasil pemekaran. Hal ini dikarenakan adanya peningkatan jumlah kehamilan pasangan keluarga usia produktif di Kabupaten Pemalang. Data Dinas Kesehatan Kabupaten Pemalang menunjukan selama tahun 2018 – 2020 jumlah ibu hamil di Kabupaten Pemalang meningkat sebesar 6%. Peningkatan kepadatan jumlah penduduk yang signifikan dapat berdampak pada pemekaran desa di Kabupaten Pemalang beberapa tahun ke depan. Berdasarkan latar belakang tersebut maka penelitian tesis ini mengkaji terkait penggerombolan desa blank spot di Kabupaten Pemalang menggunakan pendekatan metode pohon keputusan, metode penskoran terboboti, metode k-means, dan metode fuzzy c-means (fcm). Tujuan kajian penelitian ini adalah menentukan metode mana yang lebih sesuai digunakan antara metode pohon keputusan dan metode penskoran terboboti (pada jenis metode supervised classification), serta antara metode k-means dan metode fuzzy c-means (pada jenis metode unsupervised classification) untuk menggerombolkan karakteristik desa blank spot di Kabupaten Pemalang. Hasil perbandingan nilai ukuran kebaikan penggerombolan pada tipe unsupervised classification, diketahui bahwa metode fuzzy c-means lebih cocok digunakan untuk menggerombolkan karakteristik desa blank spot dibandingkan metode k-means. Model penggerombolan fuzzy c-means mampu memberikan nilai koefisien silhouette sebesar 0.991, yang nilainya lebih besar dibandingkan koefisien silhouette metode k-means. Sedangkan hasil perbandingan nilai ukuran kebaikan penggerombolan pada tipe supervised classification, diketahui bahwa metode pohon keputusan memiliki nilai presisi dan sensitivity yang lebih besar dibandingkan metode penskoran terboboti. Hal tersebut menunjukan bahwa metode pohon keputusan lebih cocok digunakan untuk menggerombolkan karakteristik desa blank spot di Kabupaten Pemalang, dibandingkan metode penskoran terboboti.

Covid-19 pandemic affects habits of people around the world. Education sector is one of the sectors affected by this change in habits. Education sector in Indonesia is also experiencing policy changes, namely the transition of policy faceto- face teaching and learning process to the distance learning process (PJJ / online learning). This is in line with the circular letter from Minister Education and Culture of Indonesia number: 36962 / MPK. A/HK/2020 dated March 17, 2020, on online learning policies during the covid-19 outbreak to reduce the spread of coronavirus. Several studies have been conducted to examine the constraints PJJ process, resulting in finding that quality of internet network is majority obstacle in PJJ process. Conditions where absence of internet network in an area is commonly called blank spot. The issue of blank spots is of particular concern to President Joko Widodo. In order to minimize the problem of blank spot, the President and Kominfo RI realized the program "Indonesia is free of signals to the corners of the country". This program involves all districts / cities in Indonesia to conduct network quality surveys in the smallest areas of the village, as well as the arrangement and refinement of communication network systems. Pemalang Regency is one of the regencies in Central Java Province that conducts network quality surveys. The network quality survey was initiated by Regional Government through Diskominfo and BPS Pemalang Regency, which included a survey of quality village network, number of residents, availability of telecommunication towers, and the state of village geography. Basically, network quality survey activities require relatively no small resources and costs. So as to conduct the efficiency of field survey activities, early detection of village blank spot status is needed based on the characteristics of blank spot villages in general. The statistical method approach that can be used to suspect a village is included in the blank spot status or not is to use the clustering method. Clustering methods can be distinguished into supervised classification methods and unsupervised classification methods. The method of supervised classification that is generally used to clustering village based on village characteristics is the decision tree method and weighted scoring method. These two methods were chosen because they have an attractive and easy to interpret display of the results cluster by user in the form of diagrams and scores. While in the unsupervised classification method that is generally used to clustering village based on village characteristics is the k-means method and fuzzy c-means (fcm) method. These two methods were chosen because they have the specificity of the characteristics to clustering village based on its characteristics firmly or vaguely (allowing one village to become a member of more than one group). A network quality survey in Pemalang Regency has been conducted, but does not demand the possibility that a new survey will be conducted on village of the expansion. This is due to an increase number of pregnancies in Pemalang Regency. Data from Pemalang District Health Office showed that during 2018 - 2020 the number of pregnant women in Pemalang Regency increased 6%. A significant increase in population density can have an impact on the expansion of villages in Pemalang Regency in the next few years. Based on this background, this thesis study examines related to the clustering blank spot villages in Pemalang Regency using approach of decision tree method, weighted scoring method, kmeans method, and fuzzy c-means (fcm) method. The purpose of this research study is to determine which method is more suitable to be used between decision tree method and weighted scoring method (in the type of supervised classification method), as well as between k-means method and fuzzy c-means method (in the type of unsupervised classification method) to clustering the characteristics of blank spot villages in Pemalang Regency. Comparison goodness value of clustering in the type unsupervised classification, it is known that the fuzzy c-means method is more suitable for use to clustering characteristics of the blank spot village than the k-means method. The fuzzy c-means model is able to give a silhouette coefficient value 0.991, which is greater than the silhouette coefficient of k-means method. While the results of the comparison goodness value of clustering in the type supervised classification, it is known that the decision tree method has a greater precision and sensitivity value than weighted scoring method. This shows that the decision tree method is more suitable to be used clustering the characteristics of blank spot villages in Pemalang Regency, than weighted scoring method.