Dana, I Made Gde Meranggi
Institut Teknologi Sepuluh Nopember

Published : 2 Documents
Articles

Found 2 Documents
Search

Pemilihan Arsitektur Terbaik pada Model Deep Learning Melalui Pendekatan Desain Eksperimen untuk Peramalan Deret Waktu Nonlinier

STATISTIKA: Forum Teori dan Aplikasi Statistika Vol 18, No 2 (2018)
Publisher : Program Studi Statistika Unisba

Show Abstract | Original Source | Check in Google Scholar | Full PDF (796.186 KB)

Abstract

Penentuan arsitektur model deep learning yang tepat merupakan hal yang sangat esensial untukmendapatkan hasil ramalan dengan tingkat kesalahan minimum. Arsitektur deep learning meliputijumlah input dan variabel apa saja yang digunakan, jumlah hidden layer, jumlah neuron pada setiaphidden layer, dan fungsi aktivasi. Pada penelitian ini dilakukan studi simulasi pada salah satu modeldeep learning, yaitu deep feedforward network, dengan berbagai kombinasi arsitektur untukmendapatkan arsitektur paling optimum. Data yang digunakan merupakan data bangkitan yangmengikuti model nonlinier Exponential Smoothing Transition Auto-regressive (ESTAR) sebanyak 1000data, di mana 900 data digunakan sebagai data training dan 100 data digunakan sebagai datatesting. Ukuran evaluasi model yang digunakan adalah root mean square error of prediction (RMSEP).Hasil empiris yang didapatkan di antaranya, pemilihan input yang tepat dapat meningkatkanakurasi peramalan, serta pemilihan fungsi aktivasi dan kedalaman arsitektur sangat diperlukanuntuk mendapatkan hasil ramalan yang semakin optimum.

Hybrid model for forecasting space-time data with calendar variation effects

TELKOMNIKA (Telecommunication Computing Electronics and Control) Vol 17, No 1: February 2019
Publisher : Universitas Ahmad Dahlan

Show Abstract | Original Source | Check in Google Scholar | Full PDF (1392.556 KB)

Abstract

The aim of this research is to propose a new hybrid model, i.e. Generalized Space-Time Autoregressive with Exogenous Variable and Neural Network (GSTARX-NN) model for forecasting space-time data with calendar variation effect. GSTARX model represented as a linear component with exogenous variable particularly an effect of calendar variation, such as Eid Fitr. Whereas, NN was a model for handling a nonlinear component. There were two studies conducted in this research, i.e. simulation studies and applications on monthly inflow and outflow currency data in Bank Indonesia at East Java region. The simulation study showed that the hybrid GSTARX-NN model could capture well the data patterns, i.e. trend, seasonal, calendar variation, and both linear and nonlinear noise series. Moreover, based on RMSE at testing dataset, the results of application study on inflow and outflow data showed that the hybrid GSTARX-NN models tend to give more accurate forecast than VARX and GSTARX models. These results in line with the third M3 forecasting competition conclusion that stated hybrid or combining models, in average, yielded better forecast than individual models.