Clustering Lahan Tanam Kentang Menggunakan Hybrid Particle Swarm Optimization Dan K-Means

Auliya, Yudha Alif (2017) Clustering Lahan Tanam Kentang Menggunakan Hybrid Particle Swarm Optimization Dan K-Means. Magister thesis, Universitas Brawijaya.

Abstract

Budidaya tanaman kentang sangat dipengaruhi oleh pemilihan lahan yang sesuai. Kriteria kesesuaian lahan untuk tanaman kentang dipengaruhi oleh faktor iklim dan karakteristik lahan. Digunakan 11 kriteria kesesuaian lahan antara lain: temperatur rata-rata, curah hujan bulan pertama, curah hujan bulan kedua hingga ketiga, curah hujan bulan keempat, kelembapan udara, drainase, tekstur tanah, kedalaman efektif, pH H2O, KTK dan kemiringan lereng. Hasil clustering berupa 4 class kesesuaian lahan yaitu sangat sesuai (S1), agak sesuai (S2), sesuai marginal (S3) dan tidak sesuai (N). Clustering lahan bertujuan untuk memberikan rekomendasi kesesuaian lahan untuk ditanami tanaman kentang. Pada penelitian ini diusulkan lima arsitektur clustering yaitu K-Means, PSO, PSO K-Means, KMeans PSO dan hybrid PSO K-Means (KCPSO). Arsitektur hybrid digunakan untuk mendapatkan hasil cluster yang lebih akurat. Pada penelitian ini digunakan pendekatan baru dengan melakukan improve metode KCPSO dengan random injection. Untuk mengukur seberapa baik solusi yang dihasilkan digunakan nilai fitness. Pada kasus clustering ini fungsi fitness dihitung menggunakan silhouette koefisien. Untuk memperoleh solusi yang optimum perlu dilakukan pengujian antara lain: pengujian arsitektur clustering, pengujian parameter dan pengujian hasil clustering. Hasil yang diperoleh dari improve KCPSO menunjukkan nilai ratarata fitness terbesar dibandingkan dengan 5 arsitektur clustering lainnya dengan nilai 0.7794. Tingkat akurasi diperoleh dengan cara membandingkan hasil clustering dengan hasil kesesuaian lahan oleh pakar yang berpedoman pada framework FAO 1976 tentang evaluasi lahan. Berdasarkan pengujian hasil clustering yang dibandingkan dengan hasil pakar diperoleh tingkat akurasi sebesar 76%.

English Abstract

The cultivation of potato plants is strongly influenced by appropriate land selection. Land suitability criteria for potato crops are influenced by climatic factors and land characteristics. Used 11 land suitability criteria include: average temperature, first month rainfall, second month to third rainfall, fourth month rainfall, humidity, drainage, soil texture, effective depth, H2O pH, CEC and slope. The result of clustering is 4 class of land suitability that is very suitable (S1), rather appropriate (S2), according to marginal (S3) and not appropriate (N). Land clustering aims to provide land suitability recommendations for planting potato crops. In this research, five clustering architectures are K-Means, PSO, PSO KMeans, K-Means PSO and PSO K-Means hybrid (KCPSO). Hybrid architecture is used to get more accurate cluster results. In this research a new approach is used to improve KCPSO method with random injection. To measure how well the resulting solution used the fitness value. In the case of clustering this fitness function is calculated using a silhouette coefficient. To obtain the optimum solution needs to be tested include: testing clustering architecture, parameter testing and clustering test results. The results obtained from KCPSO boost show the greatest average fitness value compared to 5 other clustering architectures with a value of 0.7794. The accuracy level was obtained by comparing the clustering results with the result of land suitability by the experts based on the FAO 1976 framework on land evaluation. Based on the testing of clustering results compared with the expert results obtained an accuracy of 76%.

Item Type:	Thesis (Magister)
Identification Number:	TES/004/35/AUL/c/2017/041705882
Uncontrolled Keywords:	CLUSTER ANALYSIS, CLUSTER ANALYSIS - COMPUTER PROGRAMS, LAND USE, HYBRID COMPUTER, K - THEORY, POTATO - FIELD EXPERIMENTS
Subjects:	000 Computer science, information and general works > 004 Computer science > 004.3 Processing modes > 004.35 Multiprocessing
Divisions:	S2/S3 > Magister Ilmu Komputer, Fakultas Ilmu Komputer
Depositing User:	Nur Cholis
Date Deposited:	02 Aug 2017 06:47
Last Modified:	16 Dec 2020 04:10
URI:	http://repository.ub.ac.id/id/eprint/957

	Text Bagian Depan.pdf Restricted to Repository staff only Download (193kB)
	Text BAB I.pdf Restricted to Repository staff only Download (69kB)
	Text BAB II.pdf Restricted to Repository staff only Download (128kB)
	Text BAB III.pdf Restricted to Repository staff only Download (340kB)
	Text BAB IV.pdf Restricted to Repository staff only Download (362kB)
	Text BAB V.pdf Restricted to Repository staff only Download (427kB)
	Text BAB VI.pdf Restricted to Repository staff only Download (36kB)
	Text Daftar Pustaka.pdf Restricted to Repository staff only Download (39kB)

Actions (login required)

View Item