Articles

PEMODELAN REGRESI 3-LEVEL DENGAN METODE ITERATIVE GENERALIZED LEAST SQUARE (IGLS) (STUDI KASUS: LAMANYA PENDIDIKAN ANAK DI KABUPATEN SEMARANG) Paramitha, Amanda Devi; Suparti, Suparti; Wuryandari, Triastuti
Jurnal Gaussian Vol 5, No 1 (2016): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

In a research, data was used often hierarchical structure. Hierarchical data is data obtained through multistage sampling from a population with independent variables can be defined within each level and dependent variable can be defined at the lowest level. One analysis that can be used for data with a hierarchical structure is a multilevel regression analysis. The purpose of this final three-level regression analyzes to establish regression models about the length of a child's education in the District of Semarang where the individual level-1 with a factor of gender, lodged at the family level-2 by a factor of the length of father's education and duration of maternal education and nesting on the environment level-3 with factor of residence, number of elementary school the large number of junior high school and the large number of high school. Parameter estimation in 3-level regression models can use several methods, one of which is a method of Iterative Generalized Least Square (IGLS). Of cases the length of education in the district of Semarang indicate that factors influencing factor is the length of father's education and the duration of the mother's education. Keywords : Hierarchical structure, multistage sampling, multilevel regression, Iterative Generalized Least Square.
PENENTUAN MODEL KEMISKINAN DI JAWA TENGAH DENGAN MULTIVARIATE GEOGRAPHICALLY WEIGHTED REGRESSION (MGWR) Saputri, Sindy; Ispriyanti, Dwi; Wuryandari, Triastuti
Jurnal Gaussian Vol 4, No 2 (2015): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

The problem of poverty is a fundamental problem faced in a number of regions in Indonesia, to determine significant indicators on poverty by taking into account the spatial variation in the province of Central Java can use multivariate models Geographically Weighted Regression (MGWR). In the model MGWR model parameter estimation is obtained by using Weighted Least Square (WLS). Selection of the optimum bandwidth using Cross Validation (CV). The study looked for the best model among MGWR with multivariate regression and create distribution maps counties and cities in the province of Central Java based variables significantly to poverty. The results of testing the suitability of the model shows that there is no influence of spatial factors on the percentage of poor and non-poor in the province of Central Java. Variables expected to affect the percentage of poor people is a variable percentage of expenditures for food, while the percentage of the non-poor is a variable percentage of expenditure on food and the percentage of heads of household education level less than SD. Based on the AIC and the MSE obtained the best model is the model MGWR with AIC value of 44.4603 and MSE 0.454.Keywords: Cross Validation, MGWR, Poverty, Weighted Least Square
PEMILIHAN CLUSTER OPTIMUM PADA FUZZY C-MEANS (STUDI KASUS: PENGELOMPOKAN KABUPATEN/KOTA DI PROVINSI JAWA TENGAH BERDASARKAN INDIKATOR INDEKS PEMBANGUNAN MANUSIA) Purnamasari, Sarita Budiyani; Yasin, Hasbi; Wuryandari, Triastuti
Jurnal Gaussian Vol 3, No 3 (2014): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Cluster analysis is a process of separating the objects into groups, so that the objects that belong to the same group are similar to each other and different from the other objects in another group. One method of clustering is Fuzzy C-Means (FCM). FCM is used because each data in a cluster determined by a degree of membership that have value between 0 and 1. This research use two kinds of distance, Manhattan and Euclidean. To determine the proper distance in clustering district / city in Central Java based on indicators of Human Development Index (HDI), we have to calculate the ratio of the standard deviation, where the smaller value indicates a better clustering. While the optimum number of groups obtained from the minimum value of Xie Beni. Variables that used in this research are the indicators of HDI in 2012 for district / city in Central Java, consists of: Life Expectancy Value (years), Literacy Rate (percent), Average Length of School (years), and Purchasing Power Parity (thousands rupiah). The results from this research are the distance that gives a better quality is Euclidean and the optimum cluster given when the number of cluster is five with the smallest value of Xie Beni is 0,50778.
ANALISIS RANCANGAN BUJUR SANGKAR GRAECO LATIN Naifular, Yuyun; Wuryandari, Triastuti; Wilandari, Yuciana
Jurnal Gaussian Vol 3, No 1 (2014): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

The design of the experiment is a test or series of tests, using both descriptive statistics and inferential statistics that aims to transform the input variables into an output which is the response of the experiment. The Graeco Latin Square Design was built to control the diversity of component units of local control experiment of three is a row, column, and Greek letters. Terms the Graeco Latin Square Design is if the rows, columns, Latin letters, and Greek letters have the same level and each Greek letter appears only once in each row, column, and Latin letter. The steps in the analysis of the test Graeco Latin Square Design to test the normality of the error, homogeneity of variance test, determine the degrees of freedom, calculating Sum of Squares and Mean Square every factor. Next calculate the value of F for test row, column, treatments Latin letter, and treatment of Greek letters, draw up a table of variance analysis, and conclude whether there is any effect on the response variance of each source. If there is impact, it is necessary to further test using the Duncan test
KAJIAN AVAILABILITAS PADA SISTEM KOMPONEN SERI C., Avida Nugraheni; Sudarno, Sudarno; Wuryandari, Triastuti
Jurnal Gaussian Vol 2, No 3 (2013): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Availability is a measure of system performance and measures the combined effect of reliability, maintenance and logistic support on the operational effectiveness of the system. Availability of series system is derived from inherent availability of system that takes effect from mean time to failure (MTTF) and mean time to repair (MTTR). Given observed time data of microcontroller consists of processor core, memory and programmable I/O peripheral in series, is measured its system availability. By simple linier regression method, the parameter estimation is determined after data distribution known, for the mean time. Processor core has Weibull distribution for failure time data with ,   and  as regression model while repair time data is lognormal distribution with ,  and regression model is . Memory has exponential failure time data with  and  as regression model while normal repair time data has  dan  and regression model is . Failure time data distribution of programmable I/O peripherals is Weibull with ,   and regression model  while lognormal repair time data has ,  and regression model is . Due to MTTF is 11364.57 hours and MTTR is 41.59 hours, processor core?s availability is 99.64%. Availability of memory is 99.87% from MTTF is 20000 hours and MTTR is 27 hours. Programmable I/O peripheral has 18773.41 hours as MTTF and MTTR is 38.67 hours that deliver availability 99.79%. The series system availability is 99.30% means the probability of system is in the state of functioning at given time is 99.30%.
PREDIKSI DATA HARGA SAHAM HARIAN MENGGUNAKAN FEED FORWARD NEURAL NETWORKS (FFNN) DENGAN PELATIHAN ALGORITMA GENETIKA (STUDI KASUS PADA HARGA SAHAM HARIAN PT. XL AXIATA TBK) Sari, Ira Puspita; Wuryandari, Triastuti; Yasin, Hasbi
Jurnal Gaussian Vol 3, No 3 (2014): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Artificial neural network (ANN) or Neural Network (NN) is an information processing system that has characteristics similar to biological neural networks. One of the ANN models have network is quite simple and can be applied to time series data prediction is Feed Forward Neural Networks (FFNN). In general, FFNN trained using Backpropagation algorithm to obtain weights, but performance will decrease and trapped in a local minimum when applied to data that have great complexity like financial data. The solution to this problem is to train FFNN using Genetic Algorithm (GA). GA is a search algorithm that is based on the mechanism of natural selection and genetics to determine the global optimum. Training FFNN using GA is a good solution but the problem is how to understand the workings of FFNN training using the GA, the determination of the combination of the probability of crossover (), number of populations, number of generations, and the size of the tournament (k) on the AG to produce predictive value approaching actual value. One possible option is to use the technique of trial-end-error by experimenting for some combination of these four parameters. Of the 64 times the application of the AG test results to train FFNN models on daily stock price data PT. XL Axiata Tbk obtained results are sufficiently accurate predictions indicated by the proximity of the target to the output of the crossover probability () 0.8, a population of 50, the number of generations 20000 and tournament size of 4 produces the testing RMSE 107.4769.  
PERHITUNGAN DAN ANALISIS PRODUK DOMESTIK REGIONAL BRUTO (PDRB) KABUPATEN/KOTA BERDASARKAN HARGA KONSTAN (STUDI KASUS BPS KABUPATEN KENDAL) Fitriani, Fitriani; Rusgiyono, Agus; Wuryandari, Triastuti
Jurnal Gaussian Vol 2, No 2 (2013): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Gross Regional Domestic Product (GRDP) is technical term that always we heard in the civil government or in the public society. According to Statistics Indonesia, GRDP is total number of added value who producting by effort unit in that domestic area. GRDP is one of economics growth indicator in the domestic area. If GRDP is higher, then people economics prosperity must be high too, and do also that opposite. GRDP contains of 2 methods, that is GRDP at Current Market Prices and GRDP at Constant Prices. In this report will discuss about GRDP at Constant Prices with GRDP the Kendal Regency at 2000 Constant Prices in 2010 for example. Arranging GRDP at Constant Prices has purpose to find out economics condition from year to year by discern the GRDP every year. The methods to arranging GRDP at Constant Prices are revaluasi, ekstrapolasi, and deflasi. After doing the accounting by Statistics Indonesia, we obtainable GRDP the Kendal Regency at Constant Prices in 2010 in million rupiahs is 5.394.079,31. And according the analysis, GRDP from 1983 to 2011 show the linear graph that has model GRDP = -986933 +  220901 (X). This model, can use to forecasting for GRDP the Kendal Regency at Constant Prices over the next years.
ANALISIS REGRESI KEGAGALAN PROPORSIONAL DARI COX PADA DATA WAKTU TUNGGU SARJANA DENGAN SENSOR TIPE I (STUDI KASUS DI FAKULTAS SAINS DAN MATEMATIKA UNIVERSITAS DIPONEGORO) Afranda, Oka; Wuryandari, Triastuti; Ispriyanti, Dwi
Jurnal Gaussian Vol 4, No 3 (2015): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

One of the goals of studying in Higher Education Institutionis to obtain a job as soon as possible. A graduate is not required to be an unemployed. In Indonesia, the average period of waiting time for undergraduate (S1) to get the first job is 0 (zero) to 9 (nine) months. There are several factors have influenced the length of an undergraduate to get a job. They are Grade Point Average (GPA), Length of Study, etc. Therefore, it is important to know the factors influencing the waiting time of undergraduates to get a job. One method that can be used is the analysis of survival. Survival analysis is the analysis of survival time data from the initial time of the study until certain events occur. One method of survival analysis is Cox Proportional Hazard Regression. It is used to determine the relationship between one or more independent variables and the dependent variable. Cases raised in this study were the factors influencing the waiting time of graduates of the Faculty of Science and Mathematics, University of Diponegoro by using Type I data censoring. The conclusions state that the factors influencing the waiting time of graduates are Organization, Department, and Gender.Keywords:        Waiting time of undergraduate, survival analysis, Cox Proportional Hazard, Regression, University of Diponegoro.
PENDUGAAN DATA HILANG PADA RANCANGAN ACAK KELOMPOK LENGKAP DENGAN ANALISIS KOVARIAN Fitri, Vina Riyana; Wuryandari, Triastuti; Safitri, Diah
Jurnal Gaussian Vol 3, No 3 (2014): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Analysis of Covariance (ANCOVA) is mostly used in the analysis of research or experimental design. ANCOVA is the combination between regression analysis and Analysis of Variance (ANOVA). ANCOVA were used because there are some concomitant variable, which is variable that difficult to control by the researchers but an impact on observed the response variable. The purpose from concomitant variable is reduces variability in the experiment. If there is missing data on Randomized Complete Block Design (RCBD) the first must be done estimating the missing data before ANCOVA done. ANCOVA on RCBD with complete data or missing data isn?t much different, if there are missing data, the degrees of freedom is reduced by the total amount of missing data and the sum of square treatment reduced by the value of the bias. Application of tensile strength of the glue experiment to the case ANCOVA on RCBD with one missing data show no effect of treatment and group by the tensile strength of the glue. For Fe toxicity experiment with two missing data are found only treatment effect to Fe texicity. Based on value from the coefficient of variance for one missing data and two missing data showed that ANCOVA is more appropriately used than ANOVA.
PERBANDINGAN ANALISIS KLASIFIKASI NASABAH KREDIT MENGGUNAKAN REGRESI LOGISTIK BINER DAN CART (CLASSIFICATION AND REGRESSION TREES) Waluyo, Agung; Mukid, Moch. Abdul; Wuryandari, Triastuti
Jurnal Gaussian Vol 4, No 2 (2015): Jurnal Gaussian
Publisher : Departemen Statistika FSM Undip

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Credit is the greatest asset managed by the bank and also the most dominant contributor to the bank's revenue. Debtor to pay credit to the bank may smoothly or jammed. There is a relationship variables that affect a debtor smoothly or jammed in paying credit. This study aims to identify the variables that affect a debtor's credit status. The variables used in this study were gender, education level, occupation, marital status and income. Analytical methods used include Binary Logistic Regression analysis and CART (classification and regression trees). Classification accuracy of the two methods will be compared. Based on the research results of binary logistic regression showed that the variables that affect the debtor's credit status is revenue with 80% classification accuracy. While the results of CART (classification and regression trees) in the form of a decision tree shows the type of work chosen as the root node spliting, with a classification accuracy of 81%. Keywords: credit status, logistic regression, CART
Co-Authors Abdul Hoyyi Abdur Rofiq, Abdur Agung Waluyo Agus Rusgiyono Akhmad Zaki Alif Hartati Alin Citra Suardi, Alin Citra Amanda Devi Paramitha, Amanda Devi Angga Saputra Desti Annisa Intan Mayasari Annisa Rahmawati Artha Ida Sri Anggriyani, Artha Ida Sri Arya Fendha Ibnu Shina Avida Nugraheni C. Ayu Hapsari Budi Utami Bellina Ayu Rinni Desriwendi Desriwendi Desy Rahmawati Ningrat, Desy Rahmawati Desy Ratnaningrum, Desy Dewi Setya Kusumawardani Di Asih I Maruddani Diah Safitri Dian Ika Pratiwi, Dian Ika Dwi Ispriyanti DWI RAHMAWATI Etik Setyowati, Etik Fahra Pracendi Astrelita, Fahra Pracendi Faiqotul Himmah Fitri Juniaty Simatupang, Fitri Juniaty Fitriani Fitriani Friska Irnas Adiyani Galuh Riani Putri, Galuh Riani Gian Kusuma Diah Tantri, Gian Kusuma Diah Hanik Rosyidah, Hanik Haryanti Novitasari Hasbi Yasin Ibnu Athoillah Ira Puspita Sari Ishlahul Kamal, Ishlahul Landong Panahatan Hutahaean Lintang Ratri Wardhani, Lintang Ratri Maralika Yundya Sari Moch. Abdul Mukid Mohamad Reza Pahlevi, Mohamad Reza Mustafid Mustafid Noor Afifah Nova Yanti Gultom, Nova Yanti Novia Dian Ariyani, Novia Dian Oka Afranda, Oka Oktaviana Prayudhani Pritha Sekar Wijayanti Puti Cresti Ekacitta Rahma Kurnia Widyawati Restu Sri Rahayu, Restu Restu Sri Rahayu, Restu Sri Rita Rahmawati Rizky Ade Putranto, Rizky Ade Rosmalia Safitri Sab?ngatun, Sab?ngatun Sarita Budiyani Purnamasari Sayekti Dewi Anggraini Sindy Saputri Sudarno Sudarno Sugito Sugito Suparti Suparti Susi Ekawati Syarah Widyaningtyas, Syarah Tarno Tarno Tatik Widiharih Tri Murda Agus Raditya Tuan Hanni Tyas Ayu Prasanti, Tyas Ayu Vina Riyana Fitri Wulan Safitri, Wulan Wulandari, Annisa Ayu Yuciana Wilandari Yuyun Naifular