The rapid growth of information and communication technology (ICT) gives a lot of alternatives in supporting human work. ICT plays an important role in teaching and learning process. A learning system that uses ICT is called E-Learning. Statistics is still considered as a spooky subject by most of students, especially the first year students. On the other hand, it is a mandatory subject. To solve this problem, a web based statistical learning system, namely RWikiStat 2.0, has been created. It is an open source system for statistical learning supported by the technology of Wiki, and R statistical software as its statistical engine. This system is the improvement of the previous one.
In a series system, the system fails if any of the components fails. When the system functions, there may exist correlation among components because they are connected within the same system. In this paper, we consider the reliability analysis of Type-I censored life tests of series systems composed of two components with bivariate log-normal lifetime distribution. The major interest is the inference on the mean lifetimes, and the reliability functions of the system as well as components. It is interesting to discover that conventional statistical inference based on the model parameters may yield unreliable results, especially through the maximum likelihood method. Alternatively, we apply the Bayesian approach focused on the likelihood function of the the mean lifetimes after reparametrization. The results are shown to be more promising via simulation study.
Changes in human microbiome are associated with many human diseases. One important problem of microbiome data analysis is to identify the environmental/biological covariates that are associated with different bacterial taxa. Taxa count data in microbiome studies are often over-dispersed and include many zeros. To account for such an over-dispersion, we propose to use an additive logistic normal multinomial regression model to associate the covariates to bacterial compositions. The model can naturally account for sampling variabilities and zero observations and also allow for a flexible covariance structure among the bacterial taxa. In order to select the relevant covariates and to estimate the corresponding regression coefficients, we propose a group l_1 penalized likelihood estimation method for variable selection and estimation. A Monte Carlo expectation-maximization (MCEM) algorithm is developed to implement the penalized likelihood estimation. We demonstrate the method using a data set that associates human gut microbiome to diet intake in order to identify the micro-nutrients that are associated with the human gut microbiome.
The calculation of the difference between two populations having different covariates would result in a bias parameter. Propensity score classification method can be applied to each object to be classified before measuring the parameters. This method is also compared to cluster analysis. The calculation of the average difference with propensity score technique is relatively better compared with the calculation method of cluster analysis. This applies both to small and large data.
This paper investigates the relative efficiency of Naive Bayes classifier (BD) to Bayes decision rule, which is the optimal classification rule. Under normality, Discriminant Analysis is also optimal. This paper investigates the relative efficiency of NB compared to Bayes decision, Quadratic and Linear discriminant function. It will be seen that the efficiency of NB depends on the correlation along with the signal-to-noise-ratio (S/N) in multiplication.
The finite order multivariate normal universal portfolio is studied in this paper with the objective reducing the implementation time and computer-memory requirements substantially. The finite-order portfolios are run on some selected stock-price data sets from the local stock exchange. The wealths achieved over a modest length of time are recorded. Empirically, the performance of the finite-order multivariate normal universal portfolio is comparable to that of the Dirichlet universal portfolio and yet requiring substantially shorter implementation time.