Subanar

THE EFFECT OF DECOMPOSITION METHOD AS DATA PREPROCESSING ON NEURAL NETWORKS MODEL FOR FORECASTING TREND AND SEASONAL TIME SERIES

Recently, one of the central topics for the neural networks (NN) community is the issue of data preprocessing on the use of NN. In this paper, we will investigate this topic particularly on the effect of Decomposition method as data processing and the use of NN for modeling effectively time series with both trend and seasonal patterns. Limited empirical studies on seasonal time series forecasting with neural networks show that some find neural networks are able to model seasonality directly and prior deseasonalization is not necessary, and others conclude just the opposite. In this research, we study particularly on the effectiveness of data preprocessing, including detrending and deseasonalization by applying Decomposition method on NN modeling and forecasting performance. We use two kinds of data, simulation and real data. Simulation data are examined on multiplicative of trend and seasonality patterns. The results are compared to those obtained from the classical time series model. Our result shows that a combination of detrending and deseasonalization by applying Decomposition method is the effective data preprocessing on the use of NN for forecasting trend and seasonal time series.

Optimisasi Portofolio Mean-VaR di bawah CAPM Transformasi Koyck dengan Volatilitas Tak Konstan dan Efek Long Memory

This paper investigates a development of single machine batch scheduling for multi items with dependent processing time. The batch scheduling problem is to determine simultaneously number of batch ( ), which item and its size allocated for each batch, and processing sequences of resulting batches. We use total actual flow time as the objective of schedule performance. The multi item batch scheduling problem could be formulated into a biner-integer nonlinear programming model because the number of batch should be in integer value, the allocation of items to resulting batch need binary values, and also there are some nonlinearity on objective function and constraint due to the dependent processing time. By applying relaxation on the decision variable of number of batch ( ) as parameter, a heuristic procedure could be applied to find solution of the single machine batch scheduling problem for multi items.

OPTIMASI BIAYA DISTRIBUSI RANTAI PASOK TIGA TINGKAT DENGAN MENGGUNAKAN ALGORITMA GENETIKA ADAPTIF DAN TERDISTRIBUSI

Abstrak Manajemen rantai pasok merupakan hal yang penting. Inti utama dari manajemen rantai pasok adalah proses distribusi. Salah satu permasalahan distribusi adalah strategi keputusan dalam menentukan pengalokasian banyaknya produk yang harus dipindahkan mulai dari tingkat manufaktur hingga ke tingkat pelanggan. Penelitian ini melakukan optimasi rantai pasok tiga tingkat mulai dari manufaktur-distributor-gosir-retail. Adapun pendekatan yang dilakukan adalah algoritma genetika adaptif dan terdistribusi. Solusi berupa alokasi banyaknya produk yang dikirim pada setiap tingkat akan dimodelkan sebagai sebuah kromosom. Parameter genetika seperti jumlah kromosom dalam populasi, probabilitas crossover dan probabilitas mutasi akan secara adaptif berubah sesuai dengan kondisi populasi pada generasi tersebut. Dalam penelitian ini digunakan 3 sub populasi yang bisa melakukan pertukaran individu setiap saat sesuai dengan probabilitas migrasi. Adapun hasil penelitian yang dilakukan 30 kali untuk setiap perpaduan nilai parameter genetika menunjukkan bahwa nilai biaya terendah yang didapatkan adalah 80,910, yang terjadi pada probabilitas crossover 0.4, probabilitas mutasi 0.1, probabilitas migrasi 0.1 dan migration rate 0.1. Hasil yang diperoleh lebih baik daripada metode stepping stone yang mendapatkan biaya sebesar 89,825. Kata kunci — manajemen rantai pasok, rantai pasok tiga tingkat, algortima genetika adaptif, algoritma genetika terdistribusi. Abstract Supply chain management is critical in business area. The main core of supply chain management is the process of distribution. One issue is the distribution of decision strategies in determining the allocation of the number of products that must be moved from the level of the manufacture to the customer level. This study take optimization of three levels distribution from manufacture-distributor-wholeshale-retailer. The approach taken is adaptive and distributed genetic algorithm. Solution in the form of allocation of the number of products delivered at each level will be modeled as a chromosome. Genetic parameters such as the number of chromosomes in the population, crossover probability and adaptive mutation probability will change adaptively according to conditions on the population of that generation. This study used 3 sub-populations that exchange individuals at any time in accordance with the probability of migration. The results of research conducted 30 times for each value of the parameter genetic fusion showed that the lowest cost value obtained is 80,910, which occurs at the crossover probability 0.4, mutation probability 0.1, the probability of migration 0.1 and migration rate 0.1. This result has shown that adaptive and distributed genetic algorithm is better than stepping stone method that obtained 89,825. Keywords — management supply chain, three level supply chain, adaptive genetic algorithm, distributed genetic algorithm.

Wavelet Decomposition for Time Series : Determining Input Model by Using mRMR Criterion

Determining the level of decomposition and coefficients used as input in the wavelet modeling for time series has become an interesting problem in recent years. In this paper, the detail and scaling coefficients that would be candidates of input determined based on the value of Mutual Information. Coefficients generated through decomposition with Maximal Overlap Discrete Wavelet Transform (MODWT) were sorted by Minimal Redundancy Maximal Relevance (mRMR) criteria, then they were performed using an input modeling that had the largest value of Mutual Information in order to obtain the predicted value and the residual of the initial (unrestricted) model. Input was then added one based on the ranking of mRMR. If additional input no longer produced a significant decrease of the residual, then process was stopped and the optimal model was obtained. This technique proposed was applied in both generated random and financial time series data. 2000 AMS classifications: 62M10, 65T60