Found 3 Documents

Model Components Selection in Bayesian Model Averaging Using Occams Window for Microarray Data Astuti, Ani Budi; Iriawan, Nur; Irhamah, irhamah; Kuswanto, Heri
Journal of Natural A Vol 1, No 2 (2014)
Publisher : Fakultas MIPA Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (614.073 KB)


Microarray is an analysis for monitoring gene expression activity simultaneously. Microarray data are generated from microarray experiments having characteristics of very few number of samples where the shape of distribution is very complex and the number of measured variables is very large. Due to this specific characteristics, it requires special method to overcome this. Bayesian Model Averaging (BMA) is a Bayesian solution method that is capable to handle microarray data with a best single model constructed by combining all possible models in which the posterior distribution of all the best models will be averaged. There are several method that can be used to select the model components in Bayesian Model Averaging (BMA). One of the method that can be used is the Occams Window method. The purpose of this study is to measure the performance of Occams Window method in the selection of the best model components in the Bayesian Model Averaging (BMA). The data used in this study are some of the gene expression data as a result of microarray experiments used in the study of Sebastiani, Xie and Ramoni in 2006. The results showed that the Occams Window method can reduce a number of models that may be formed as much as 65.7% so that the formation of a single model with Bayesian Model Averaging method (BMA) only involves the model of 34.3%. Keywords— Bayesian Model Averaging, Microarray Data, Model Components Selection, Occams Window Method.
STATISTIKA: Forum Teori dan Aplikasi Statistika Vol 3, No 1 (2003)
Publisher : Program Studi Statistika Unisba

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29313/jstat.v3i1.532


Analisis korelasi kanonik adalah analisis statistik multivariate yang bertujuan untuk memeriksa adanya hubunganantara dua kelompok variabel dengan cara memaksimumkan nilai korelasi antar variabel baru yang merupakan kombinasi lineardari variabel yang ada pada setiap kelompok. Nilai korelasi tersebut rentan terhadap adanya pengamatan yang berpengaruh(influence-point). Untuk mengatasi kerentanan tersebut, diusulkan penggunaan metode robust. Dalam makalah ini disajikanperbandingan antara penggunaan metode robust dengan metode non-robust pada data kependudukan negara-negara di Asia.Metode non-robust yang digunakan adalah metode minimumcovariance determinant. Dari hasil penelitian ini, ada beberapanegara Asia yang dapat dianggap sebagai influence-point dan terdapat perbedaan hasil antara metode robust korelasi kanonik dannon-robust korelasikanonik.
FUZZY T2 HOTELLING (T_F^2 ) CONTROL CHART Kesumawati, Ayundyah; Mashuri, Mashuri; Irhamah, Irhamah
EKSAKTA: Journal of Sciences and Data Analysis VOLUME 14, ISSUE 1, February 2014
Publisher : Fakultas Matematika dan Ilmu Pengetahuan Alam

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


Statistical Process Control (SPC) is a method used to monitor a process for identifying special causes of variation and necessary to improve the process. One technique commonly used in the SPC is to determine whether the process is stable or not, both the mean and variability. Multivariate control charts are used if necessary to control together two or more related quality characteristics. Sometimes in a process production there is a lack of precision in the calculation, especially if the data used in the form of either data or qualitative attributes. Fuzzy set theory, specifically discusses the development of concepts and techniques related to the sources of uncertainty or imprecision in nature. Control charts are constructed by transforming crisp numbers into fuzzy numbers can be an alternative to obtain representative results of several variables in which there are several quality characteristics. Transformation of some functions, which are used in this study is Fuzzy Median Transformation (FMT). The advantages of FMT is that it can be used for the data in the form of asymmetry. This paper will discuss about the algorithm for  Fuzzy T2  Hotelling control chart and its application to the production data of PT. IGLAS (Persero). From the results of the application of Fuzzy T2  Hotelling control chart got that out of the 5 variables that were analyzed, the dominant variables that lead to out of control is variable bottle molding process