FORUM STATISTIKA DAN KOMPUTASI
Not a member yet
119 research outputs found
Sort by
ADVECTION-DIFFUSION MODEL WITH TIME DEPENDENT FOR AIR POLLUTANTS DISTRIBUTION IN UNSTABLE ATMOSPHERIC CONDITION
Full text linkAir pollution levels are quite high in urban areas. They are emitted from various sources and have an impact on humans and the environment. There are some physical processes that occur when pollutants disperse in the atmosphere. The main processes are advection and diffusion. Therefore, a two-dimensional mathematical model is presented to study the dispersion of air pollution under the effect of mesoscale wind as an effect of urban heat islands. This model is solved by using the implicit Crank-Nicolson finite difference scheme under stability-dependent meteorological parameters involved in large scale wind, mesoscale wind and eddy diffusivity. The main goal of this research is to analyze air pollution distribution using the advection-diffusion model. The results of this model have been analyzed for the dispersion of air pollutants in an urban area in the downwind and vertical direction for unstable atmospheric conditions.Key words : Advection, Diffusion, Mesoscale Wind, Pollutant Dispersio
NONLINEAR PRINCIPAL COMPONENT ANALYSIS AND PRINCIPAL COMPONENT ANALYSIS WITH SUCCESSIVE INTERVAL IN K-MEANS CLUSTER ANALYSIS
Full text linkK-Means Cluster is a cluster analysis for continuous variables with the concept of distance used is a euclidean distance where that distance is used as observation variables which are uncorrelated with each other. The case with the type data that is correlated categorical can be solved either by Nonlinear Principal Component Analysis or by making categorical data into numerical data by the method called successive interval and then used Principal Component Analysis. By comparing the ratio of the variance within cluster and between cluster in poverty data of East Nusa Tenggara Province in K-Means cluster obtained that Principal Component Analysis with Successive interval has a smaller variance ratio than Nonlinear Principal Component Analysis. Variables that take effect to the clusterformation are toilet, fuel,and job.Keywords: K-Means Cluster Analysis, Nonlinear Principal Component Analysis, Principal Component Analysis, Successive interval
PENDEKATAN KUADRAT TERKECIL PARSIAL KEKAR UNTUK PENANGANAN PENCILAN PADA DATA KALIBRASI
Full text linkThe serious problems in the calibration of multivariate estimation are multicollinearity and outliers. Partial Least Squares (PLS) is one of the statistical method used in chemometrics, to handle high or perfect multicollinearity in independent variables. Straightforward Implementation Partial Least Squares (SIMPLS) is the extension of PLS regression proposed by De Jong (1993). The SIMPLS algorithm is based on the empirical cross-variance matrix between the independent variables and the regressors. This method does not resistant toward outlier observations. Robust PLS method is used to handle the multicollinearity and outliers in the data sets. This method can be classified in two groups, there are iteratively reweighting technique and robustication of covariance matrix. Partial Regression-M (PRM) method is one of the robust PLS methods used the idea of iteratively reweighting technique that proposed by Serneels et al. (2005). Robust SIMPLS (RSIMPLS) method is one of the robust PLS methods used the idea of robustication of covariance that proposed by Huber and Branden (2003). A modified RSIMPLS used M estimator with the Huber weight function called RSIMPLS-M was proposed by Ismah (2010). These two methods (RSIMPLS-M and PRM) are applied to Fish data (Naes 1985) to know their performances. The research results indicated that the values of R2 and RMSEP of RSIMPLS-M are higher than those of PRM method. Whereas based on the confidence interval estimation of the regression coefficients by jackknife method, estimation of PRM is narrower than that RSIMPLS-M method. Therefore RSIMPLS-M method is better than PRM method for prediction, whereas PRM method is better than RSIMPLS-M method for estimation.Keywords : Partial least squares regression robust (PLSRR), partial robust M-regression (PRM), straightforward implementation partial least squares robust (RSIMPLS
DETEKSI DINI RISIKO KREDIT MELALUI RATING TRANSITION STOCHASTIC MATRIX DAN VALUE AT RISK (Early Detection of Credit Risk Through Rating Transition Stochastic Matrix and Value at Risk)
Full text linkCredit risk is the risk occurs when the debtors fail to meet their obligation in accordance with agreed term to the bank. This research is made to analyze the credit risk for industrial and trade sector in Bank X, both sectors contribute about 80% loan credit. The calculation of the VaR 95% used Markov Chain regular and ergodic and adjusted by macro economic variable which significance influence the movement of those quality rating. The result of Markov chain for industrial sector show that the ability debtor increase for repay the loan in the long run but for trade sector became worst. The VaR 95% results for industrial sector is Rp 2,17 billion or about 3,27% and for trade sector is Rp 4,46 billion or about 2,03% from outstanding credit those sectors. This results is not appropriate with the New Basel Capital Accord which recomennded to allocate capital 8% from outstanding credit to cover credit risk. The calculation of the TVaR 95% for industrial sector is Rp 4,89 billion or about 7,38% and for trade sector is Rp 16,60 billion or about 7,55% from outstanding credit both sectors. For the TVaR 95% portofolio give the results is Rp 18,99 billion or about 6,5% from outstanding credit.Keywords : Credit Risk, Markov chain, Regression, Macroeconomics, VaR, TVaR, Portofolio Risk
Penerapan Multivariate Cusum Time Series untuk Mendeteksi Kegagalan Bank di Indonesia
Full text linkBank merniliki peran penting dalam pengalokasian sumberdaya keuangan. Kondisi bank yangtidak sehat dapat menyebabkan bank tidak dapat menjalatzkan peran tersebut, sehingga akanmenghanlbat kelancaran akt$tas perekonomian nasional. Dalam mengevaluasi kinerja bank,beberapa pendekatan metodologi terutama metodologi statistik telah banyak dilakukan. Nalnunselama ini nzetodologi tersebut tidak mengikutsertakan perilaku deret waktu dari peubah-;7eubahnya. Padahal peubah-peubah keuangan suatu perusahaan secara serial berkorelasi tinggi.Tulisan ini bertujuan untuk mendeteksi kegagalan bank dengan menggunakan multivariatecztsunz tiine series.Model kegagalan bank yang dibangun oleh multivariate clrsunz time series, cukup mampu dalanzrnendeteksi adanya gejala memburuk pada kondisi kesehatan bank. Hal ini sejalan dengansenzangat pendeteksian krisis perbankan secara dini (early warning banking crises).Kata kunci : Multivariate Cusum Time Series, Kegagalan Ban
BOOTSTRAP PARAMETRIK DAN NONPARAMETRIK UNTUK PENDUGAAN KUADRAT TENGAH GALAT DALAM STATISTIK AREA KECIL DENGAN RESPON BERSEBARAN LOGNORMAL
Full text linkSmall area estimation is needed to obtain information in small area, that is area containing small size of sample. Direct estimation in small area will result in large variance. Indirect estimation is the solution, with involves auxiliary data from related area or another survey in parameter estimation. One of the prolems found in using this procedure is low precision of Mean Square Error (MSE) estimate caused by non-normal distribution. Parameter of concern in this study is per capita expenditure of village in Bogor regency. Per capita expenditure is non-normal distribution. MSE estimator with bootstrap method has the advantage of potential robustness against sampling from non- normal distribution. Therefore this study used bootstrap method, such as parametric bootstrap and nonparametric bootstrap, in MSE estimation. Generally, the result showed that the MSE estimate of the parametric bootstrap smaller than the nonparametric bootstrap. Both method have better precision, so that they can repair the result of direct estimation.Keywords : small area estimation, parametric bootstrap, nonparametric bootstra
SMALL AREA ESTIMATION OF LITERACY RATES ON SUB-DISTRICT LEVEL IN DISTRICT OF DONGGALA WITH HIERARCHICAL BAYES METHOD
Full text linkLiteracy Rate (LR) is defined as percentage of population aged over 15 with ability to read and write. LR, as one of people welfare indicators, is a measurement of educational development. The indicator, as a measurement of government performance on education, can be measured if all variables related is available. Statistics Indonesia (BPS) each year calculated LR based on National Socio-Economic Survey (SUSENAS) with estimation available only on provincial level and district level. Along with establishment of autonomous regional policy, where regional government had greater power to manage its own region, availability of LR on lower levels to monitor educational development is necessary. Due to sampling design of SUSENAS, accommodated only estimation on district level, will give high variance if used to estimate on lower sub-district level, although still unbiased. Modelling LR was done with Logit-Normal approach, because LR data followed Binomial Distribution. Good estimators from inadequate sample size can be obtained with method of Small Area Estimation (SAE). Hierarchical Bayes (HB) method is one of SAE methods which are proven to give good estimate on binomial distributed data as LR. Estimation on sub-district level in District of Donggala with HB method gave better result compared to the direct estimation with lower Mean Square Error (MSE).Key words : Small Area Estimation, Literacy Rate, Hierarchical Bayes, Logit-Normal Mode
MODELLING OF FORECASTING MONTHLY INFLATION BY USING VARIMA AND GSTARIMA MODELS
Full text linkThe model parameters could be different form the well to the factors of time and location. A general model of GSTAR can be used to establish model the inflation in some locations by using GSTARIMA model if time series data is self-contained autoregressive, differentiation, and moving averages. This study examines whether the effect of such locations on the GSTARIMA model is better than the VARIMA model that regardless of the location influences. The aim of this study is to establish two models of inflation six provincial capitals in Java using VARIMA model and GSTARIMA model with inverse distance weighting. Dummy variables have been used to overcome normality and white noise problems. The best forecasting of monthly inflation in provincial captitals in Java Island is GSTAR(1;1) with inverse distance weighting. It has smallest RMSE value of 0.9199.Key words : GSTARIMA, Inverse Distance, RMSE, VARIM
ALTERNATIVE SEMIPARAMETRIC ESTIMATION FOR NON-NORMALITY IN CENSORED REGRESSION MODEL WITH LARGE NUMBER OF ZERO OBSERVATION
Full text linkA large number of zero observation on the response variable in the socio-economic field are often found in household demand models. This will imply on the method to estimate parameters in the model used. Ordinary least square estimators of linear models to be biased and inconsistent. One model to overcome is using censored regression model is also know as tobit model. However, non-normality in the Tobit Estimators being inconsistent. Another alternative estimators is censor least absolute deviations (CLAD). CLAD estimator is consistent and asymptotically normal for a wide class of distribution. This study was to focus on the application of Tobit and Censored Least Absolute Deviations (CLAD) estimators for LPG demand. The data used is the LPG expenditure in rural areas in the provinces of West Java that the number zero observations is 39 percent of the sample. The result shows that CLAD and Tobit estimators are consistent estimators. But along with increasing the number of samples, the CLAD estimators performance is getting better than Tobit estimators.Keywords : Zero observation, CLAD, Tobit, Consistent estimator, LPG deman
RIDGE AND LASSO PERFORMANCE IN SPATIAL DATA WITH HETEROGENEITY AND MULTICOLLINEARITY
Full text linkSpatial heterogeneity becomes a separate issue on the analysis of spatial data. GWR (Geographically Weighted Regression) is a statistical technique to explore spatial nonstationarity by form the differrent regression models at different point in observation space. Multicollinearity is a condition that the independent variables in model have linear relationship. It would be a problem for estimation parameters process, because that condition produces unstable model. This problem may be found in GWR models, which allow the linear relationship between independent variables at each location called local multicollinearity. GWRR (Geographically Weighted Ridge Regression) and GWL (Geographically Weighted Lasso) which use the concept of ridge and lasso is shrink the regression coefficient in GWR model. GWRR and GWL techniques are consider to be capable of overcoming local multicollinearity to produce more stable models with lower variance. In this study, GWRR and GWL is used to model Gross Regional Domestic Product (GRDP) in Java using kernel exponential weighted function. The results showed that GWL has better performance to predict GRDP with lower RMSE and higher value than GWRR.Keyword : Spatial Heterogeneity, GWR, Local Multicollinearity, Ridge, Lass