Bambang Widjanarko Otok

Ensemble method based on ARIMA-FFNN for climate forecasting

Ensemble forecasting is one of relatively new modern methods for time series forecasting that employs averaging or stacking from the results of several methods. This paper focuses on the development of ensemble ARIMA-FFNN for climate forecasting by using averaging method. Two data about monthly rainfall in Indonesia, i.e. Wagir and Pujon region, are used as case study. Root mean of squares errors in training and testing datasets are used for evaluating the forecast accuracy. The results of ensemble ARIMA-FFNN are compared to one classical statistical method, i.e. individual ARIMA, and two modern statistical methods, namely individual FFNN and ensemble FFNN. The results show that ARIMA yields more accurate forecast in training datasets than other methods, whereas in testing datasets show that FFNN is the best method. Additionally, this conclusion in line with the results of M3 competition, i.e. modern methods or complex methods do not necessarily produce more accurate forecast than simpler one.

Design of Experiment to Optimize the Architecture of Wavelet Neural Network for Forecasting the Tourist Arrivals in Indonesia

Wavelet Neural Network (WNN) is a method based on the combination of neural network and wavelet theories. The disadvantage of WNN is the lack of structured method to determine the optimum level of WNN factors, which are mostly set by trial and error. The factors affecting the performance of WNN are the level of MODWT decomposition, the wavelet family, the lag inputs, and the number of neurons in the hidden layer. This research presents the use of design of experiments for planning the possible combination of factor levels in order to get the best WNN. The number of tourist arrivals in Indonesia via Soekarno-Hatta airport in Jakarta and via Ngurah Rai airport in Bali is used as case study. The result shows that design of experiments is a practical approach to determine the best combination of WNN factor level. The best WNN for data in Soekarno-Hatta airport is WNN with level 4 of MODWT decomposition, Daubechies wavelet, and 1 neuron in the hidden layer. Whereas, the best WNN for data in Ngurah Rai airport is WNN with MODWT decomposition level 3 and using input proposed by Renaud, Starck, and Murtagh [11] and seasonal lag input addition.

Propensity score matching of the gymnastics for diabetes mellitus using logistic regression

Diabetes Mellitus (DM) is a group of metabolic diseases with characteristics shows an abnormal blood glucose level occurring due to pancreatic insulin deficiency, decreased insulin effectiveness or both. The report from the ministry of health shows that DMs prevalence data of East Java province is 2.1%, while the DMs prevalence of Indonesia is only 1,5%. Given the high cases of DM in East Java, it needs the preventive action to control factors causing the complication of DM. This study aims to determine the combination factors causing the complication of DM to reduce the bias by confounding variables using Propensity Score Matching (PSM) with the method of propensity score estimation is binary logistic regression. The data used in this study is the medical record from As-Shafa clinic consisting of 6 covariates and health complication as response variable. The result of PSM analysis showed that there are 22 of 126 DMs patients attending gymnastics paired with patients who didnt attend to diabetes gymnastic...

Hierarchical linear models on abdominal circumference research data using generalized least square

Hierarchical Linear Models (HLM) is one of multilevel statistical analysis which is the development of a linear regression analysis on individual data, where data structured hierarchical (tiered). The dependent variable was measured at level-1 or at the lowest level only, whereas the independent variables measured at level-1 and level higher. In this study will use data from Riskesdas and Susenas on 2013 in the province of East Java. With the unit analysis in level 1 is 54.101 individuals and the unit analysis in level-2 is 38 regencies / cities in East Java. The data related to the obesity which is a condition of abnormal or excess accumulation of fat in adipose tissue. Physical activity affects the central obesity, especially abdominal circumference. Consumption of fruit and vegetables are also thought to affect the abdominal circumference. To improve the health status of the community, needs to be studied in-depth the factors that affect the abdominal circumference. This study is using Hierarchical Lin...

Path, centroid, and factor scheme for modeling the remuneration of educational staff in ITS with partial least square (PLS)

Bureaucracy condition in Indonesia reveals many shortcomings. One of bureaucratic reformation from the government is the remuneration for Civil Servants (PNS). Remuneration is a part of welfare received by employees, which can be used as an element of motivation for employees to excel and improve their performance. Variables in this study are interrelated. Motivation for achievement (ξ1), characteristics of work environment (ξ2) and training transfer (ξ3) are supposedly affect the performance (η1), while the performance (η1) affects the remuneration (η2). Both the performance and remuneration are constructs or latent variables, which cannot be measured directly. Therefore, the SEM method is considered able to resolve these problems. However, SEM has some assumptions that must be met. The assumptions were frequently violated when real data is used, so we need a method that is free of assumptions, free distribution and flexible that is variance-based SEM or namely partial least square (PLS). PLS is an estim...

Bootstrap inference longitudinal semiparametric regression model

Semiparametric regression contains two components, i.e. parametric and nonparametric component. Semiparametric regression model is represented by yti=μ(x˜′ti,zti)+eti where μ(x˜′ti,zti)=x˜′tiβ˜+g(zti) and yti is response variable. It is assumed to have a linear relationship with the predictor variables x˜′ti=(x1i1,x2i2,…,xTir). Random error eti, i = 1, …, n, t = 1, …, T is normally distributed with zero mean and variance σ2 and g(zti) is a nonparametric component. The results of this study showed that the PLS approach on longitudinal semiparametric regression models obtain estimators β˜^t=[X′H(λ)X]−1X′H(λ)y˜ and g˜^λ(z)=M(λ)y˜. The result also show that bootstrap was valid on longitudinal semiparametric regression model with g^λ(b)(z) as nonparametric component estimator.

Simulation study for biresponses nonparametric regression model using MARS

In statistical modeling, especially regression analysis, we can find relationship pattern between two responses with several predictors and both of responses are correlated each other. When the assumption of the pattern is unknown, then the regression parameters could be obtained by using biresponses nonparametric regression. One method that often used in nonparametric regression with single response is Multivariate Adaptive Regression Spline (MARS). This paper aims to know how ability of MARS in estimating biresponses nonparametric regression through simulation study on different sample size (n) and variance error (σ2). We use R-square and MSE as the goodness of fit criterion. Result shows that the smaller variance error gives better estimation than the bigger one, because it gives higher R-square and smaller MSE values. Whereas the variation of sample size gives small effect on the accuracy of the model, because the value of R-square and MSE in this case tend to be the same on different sample sizes.

Bootstrap Aggregating Multivariate Adaptive Regression Splines (Bagging MARS) for Poor Households Classification in Region of Jombang

Poverty is a major social problems of each country, especially developing countries, including Indonesia. The focus of this research is poverty in Jombang because of an increase Human Development Index and the economy is not accompanied by a decrease in the poverty rate in Jombang from 2009 until 2011, this is presumably because the provision of assistance to poor households in Jombang not on target, so needed classification method to help the poor households for assistance can be precisely targeted. MARS is a classification method that is focused to overcome the problems of high dimensions and discontinuities in the data. The accuracy of classification can be improved using the resampling methods, namely bagging. This research will use bagging MARS method to obtain classification models based aid poor households are expected in Jombang. MARS modeling results concluded that the probability of poor households in Jombang who need primary assistance is 0.789 and probability of secondary needs help is 0.211, and there are 14 variables that affect the expected help poor households in Jombang. The accuracy of MARS method for classification is 69.40 percent, while the accuracy of the bagging MARS method is 69.63 percent, this accuracy is the best among 25, 50, 100, 150, 200, 250, and 500 replication. Thus, in this research, a more precise method used to classify the expected help poor households in Jombang is bagging MARS.