Seleksi Fitur Menggunakan Backward Elimination Pada Prediksi Cuaca Dengan Neural Network
Sari
Abstrak
Prakiraan Cuaca merupakan bagian yang penting untuk pengambilan keputusan seperti kontrol lalu lintas udara, kapal routing, pertanian, pembangkit listrik dan keuangan. Oleh karena itu banyak pihak yang membutuhkan informasi kondisi atmosfer (cuaca) yang lebih cepat, lengkap, dan akurat. Metode Neural Network lebih efisien dalam perhitungan cepat dan mampu menangani data yang tidak stabil yang khas dalam kasus data prakiraan cuaca , tetapi Neural Network menghadapi keterbatasan dalam mempelajari pola klasifikasi jika dataset memiliki data yang luar biasa dan dimensi yang kompleks. Untuk menyelesaikan masalah tersebut diperlukan metode penyeleksian fitur. Metode yang digunakan adalah Backward Elimination Untuk Seleksi Fitur Pada Metode Neural Network. Untuk Prediksi Cuaca dengan input data adalah data sinoptik. Beberapa percobaan dilakukan untuk mendapatkan arsitektur yang optimal dan menghasilkan prediksi yang akurat. Hasil penelitian menunjukkan metode jaringan syaraf tiruan berbasis backward elimination menghasilkan peningkatan akurasi 4% dibandingkan hanya dengan menggunakan metode jaringan syaraf tiruan saja. Melalui penelitan ini dapat diketahui sejauh mana metode Neural Network menggunakan seleksi fitur Backward Elimination mempunyai tingkat akurasi lebih baik sehingga dapat membantu banyak pihak yang membutuhkan informasi kondisi atmosfer (cuaca) yang lebih cepat, lengkap, dan akurat.
Kata Kunci: prediksi, cuaca, Jaringan Syaraf Tiruan, Backward Elimination. AbstractWeather Forecast is an important part of the decision making such as air traffic control, ship routing, agriculture, power generation and financial. Therefore, many people who need information atmospheric conditions (weather) is more rapid, complete, and accurate. Neural Network method is more efficient in calculations quickly and is able to handle data that is not stable in the case of typical weather forecast data, but Neural Network face limitations in studying the pattern classification if the dataset has exceptional data and complex dimensions. To resolve these problems required method of selecting features. The method used is the Backward Elimination for Seleksi Fitur Method of Neural Network On. For weather prediction with the data input is data synoptic. Several experiments were conducted to obtain the optimal architecture and generate accurate predictions. The results show the method of artificial Neural Network-based backward elimination resulted in increased accuracy compared to only 4% using Neural Network. Through this research can be known to what extent the method of Neural Network using Backward Elimination feature selection has a better accuracy so that it can help many parties who require information of the condition of the atmosphere (weather) faster, complete, and accurate.Keywords: Prediction, Weather, Neural Network, Backward EliminationTeks Lengkap:
PDFReferensi
Aldrian, E., & Djamil, Y. S. (2008). Application of Multivariate ANFIS For Daily Rainfall Prediction : Influences of Training Data Size. MAKARA, 12(April 2008), 7–14.
Beniwal, S., & Arora, J. (2012). Classification and Feature Selection Techniques in Data Mining. International Journal of Engineering Research & Technology (IJERT), 1(6), 1–6.
Chris Fraley, Adrian Raftery, Tilmann Gneeting, McLean Sloughter & Veronica Berrocal. Probabilistic Weather Forecasting in R. The R Journal. 56-62.
Felkin, M. (2007). between N -ary and Binary Problems. Between N-ary and Binary Problems, 1–25.
Han & Kamber. (2006). Data Mining: Concept and Technique (2nd ed). United State America. Heaton. (2008). Introduction to Neural Network With Java (2nd ed). USA. Heaton Research,Inc.
Hong, X., Harris, C., Brown, M., & Chen, S. (2002). Backward Elimination Methods for Associative Memory Network Pruning. Computers and Technology, (Reed 1993).
Hung, N. Q., Babel, M. S., Weesakul, S., & Tripathi, N. K. (2009). An artificial neural network model for rainfall forecasting in Bangkok , Thailand. Hydrolgogy and Earth System Sciences, 1413–1425.
Jong Jek Siang. (2009). Jaringan Syaraf Tiruan & Pemrogramannya Menggunakan MATLAB. Yogyakarta : Andi Yogyakarta.
Kohavi, R., & John, H. (1997). Artificial Intelligence Wrappers for feature subset selection. elsevier, 97(97), 273–324.
Kusumadewi, Sri & Hartati, Sri. (2010). Neuro-Fuzzy Integrasi Sistem Fuzzy & Jaringan Syaraf. (2nd ed). Yogyakarta: Graha Ilmu.
Lin, S.-W., Chen, S.-C., Wu, W.-J., & Chen, C.-H. (2009). Parameter determination and feature selection for back-propagation network by particle swarm optimization. Knowledge and Information Systems, 21(2), 249–266.
Liu, H., Member, S., Yu, L., & Member, S. (2005). Algorithms for Classification and Clustering. IEEE Transaction on Knowledge and Data Engineering, 17(April 2005), 491–502. Program Pascasarjana Magister Ilmu Komputer STMIK Nusa Mandiri
Pallavi, V. P., Vaithiyanathan, V., & Ph, D. (2013). Combined Artificial Neural Network and Genetic Algorithm for Cloud Classification. International Journal of Engineering Research & Technology (IJET), (May), 787–794.
Radhika, Y., & Shashi, M. (2009). Atmospheric Temperature Prediction using Support Vector Machines. International Jurnal Of Computer Theory and Engineering, 1(April), 55–58. doi:10.7763/IJCTE.2009.V1.9
Selection, O. F., & Elimination, B. (2004). Orthogonal Forward Selection and Backward Elimination for Feature Subset Selection. IEEE Transactions on System Man and Cybernetics, 1(February), 629–634.
Solaimani, K. (2009). Rainfall-runoff Prediction Based on Artificial Neural Network ( A Case Study : Jarahi Watershed ). IDOSI Publication, 5(6), 856– 865. Vercellis, C. (2009). Business Intelligence: Data Mining and Optimization for Decision Making.
Wiley. Winarso.(2002). Pemikiran dan Praktek Perencanaan dalam Era Tranformasi di Indonesia. Bandung: Departemen Teknik Planologi ITB.
Zhu, Z., Ong, Y., & Dash, M. (n.d.). Wrapper-Filter Feature Selection Algorithm Using A Memetic Framework. International Journal of Engineering Research & Technology (IJET), 1–19
DOI: https://doi.org/10.31294/ijcit.v2i1.1911
##submission.copyrightStatement##
##submission.license.cc.by-sa4.footer##
P-ISSN: 2527-449X E-ISSN: 2549-7421
Statistik Pengunjung Jurnal IJCIT
