PERBANDINGAN METODE ARIMA (BOX-JENKINS) DENGAN JARINGAN SARAF TIRUAN (JST) BACKPROPAGATION SEBAGAI METODE PERAMALAN RATA-RATA TEMPERATUR BUMI

• Main Author: Sihombing, Oksendi Vitra (Author)
• Format: Book
• Published: 2013-11-08.
• Subjects: Thesis; NonPeerReviewed
• Online Access: Link Metadata

Summary

Temperatur rata-rata bumi setiap tahun dalam dekade terakhir ini terus meningkat dibanding beberapa dekade lalu. Rata-rata temperatur bumi telah naik sekitar 0,8oC sejak 1880. Kenaikan emisi Gas Rumah Kaca (GRK) di atmosfer akan menyebabkan kenaikan temperatur bumi dalam jangka waktu panjang. Dampak yang ditimbulkan akibat kenaikan rata-rata temperatur bumi adalah kekeringan, krisis air, hingga perubahan cuaca secara global. Untuk mengantisipasi perubahan rata-rata temperatur secara ekstrim diperlukan suatu model yang dapat meramalkan kondisi temperatur. Model yang banyak digunakan adalah model ARIMA untuk pendekatan model linear. Berdasarkan proses identifikasi model ARIMA, model yang didapat adalah Model ARIMA (3,1,0). Sedangkan untuk model pendugaan data rata-rata temperatur bumi dengan menggunakan Jaringan Saraf Tiruan (JST) dengan metode backpropagation menghasilkan model optimum BPNN (4,10,5,1). Berdasarkan kedua model tersebut yang memberikan nilai MAPE terkecil adalah BPNN (4,10,5,1) yaitu sebesar 0,003988183 % dibandingkan model ARIMA (3,1,0) sebesar 0,00498963 %. Kata kunci : Temperatur, Peramalan, ARIMA, Box-Jenkins, Jaringan Saraf Tiruan, JST, Backpropagation. The earth's average temperature every year in the last decade has steadily increased over the past few decades. Earth's average temperature has increased by about 0.8 ° C since 1880. The increase of Greenhouse Gas emissions (GHG) in the atmosphere will cause the earth's temperature rises in a long term. Impacts caused by the increase of the earth's average temperature is a drought, water crisis, to global climate changes. To anticipate the extreme changes in the average temperature, a model that can predict temperature conditions is required. The model that widely used is ARIMA for linear model approach. Based on the ARIMA model identification process, ARIMA (3,1,0) is obtained. As for the prediction model for the earth's average temperature data by using Artificial Neural Network (ANN) with the Backpropagation method produces BPNN (4,10,5,1) as the optimal model. Based on these two models, the model which gives the smallest MAPE value is BPNN (4,10,5,1) that is equal to 0.003988183% compared ARIMA (3,1,0) of 0.00498963%. Keywords : Temperature, Forecasting, ARIMA, Box-Jenkins, Artificial Neural Networks, ANN, Backpropagation.

Item Description

http://repository.upi.edu/2943/1/S_MTK_0907081_TITLE.pdf
http://repository.upi.edu/2943/2/S_MTK_0907081_ABSTRACT.pdf
http://repository.upi.edu/2943/3/S_MTK_0907081_TABLE%20OF%20CONTENT.pdf
http://repository.upi.edu/2943/4/S_MTK_0907081_CHAPTER1.pdf
http://repository.upi.edu/2943/5/S_MTK_0907081_CHAPTER2.pdf
http://repository.upi.edu/2943/6/S_MTK_0907081_CHAPTER3.pdf
http://repository.upi.edu/2943/7/S_MTK_0907081_CHAPTER4.pdf
http://repository.upi.edu/2943/8/S_MTK_0907081_CHAPTER5.pdf
http://repository.upi.edu/2943/9/S_MTK_0907081_BIBLIOGRAPHY.pdf
http://repository.upi.edu/2943/10/S_MTK_0907081_APPENDIX.pdf