Tai Vovan

Statistical Discrimination Analysis Using the Maximum Function

The maximum of k functions defined on R n , n ≥ 1, by f max (x) = max{f 1 (x),…, f k (x)}, ∀ x ∈ R n , can have important roles in Statistics, particularly in Classification. Through its relation with the Bayes error, which is the reference error in classification, it can serve to compute numerical bounds for errors in other classification schemes. It can also serve to define the joint L1-distance between more than two densities, which, in turn, will serve as a useful tool in Classification and Cluster Analyses. It has a vast potential application in digital image processing too. Finally, its versatile role can be seen in several numerical examples, related to the analysis of Fishers classical iris data in multidimensional spaces.

Fuzzy clustering of probability density functions

ABSTRACT Basing on L1-distance and representing element of cluster, the article proposes new three algorithms in Fuzzy Clustering of probability density Functions (FCF). They are hierarchical approach, non-hierarchical approach and the algorithm to determine the optimal number of clusters and the initial partition matrix to improve the qualities of established clusters in non-hierarchical approach. With proposed algorithms, FCF has more advantageous than Non-fuzzy Clustering of probability density Functions. These algorithms are applied for recognizing images from Texture and Corel database and practical problem about studying and training marks of students at an university. Many Matlab programs are established for computation in proposed algorithms. These programs are not only used to compute effectively the numerical examples of this article but also to be applied for many different realistic problems.

Modified genetic algorithm-based clustering for probability density functions

ABSTRACT This article modifies two internal validity measures and applies them to evaluate the quality of clustering for probability density functions (pdfs). Based on these measures, we propose a new modified genetic algorithm called GA-CDF to establish the suitable clusters for pdfs. The proposed algorithm is tested by four numerical examples including two synthetic data sets and two real data sets. These examples illustrate the superiority of proposed algorithm over some existing algorithms in evaluating the internal or external validity measures. It demonstrates the feasibility and applicability of the GA-CDF for practical problems in data mining.

An Improved Fuzzy Time Series Forecasting Model

This model is developed from the model of Abbasov and Mamedova (2003) in which the parameters are investigated by methods and algorithm to obtain the most suitable values for each data set. The experiments on Azerbaijan’s population, Vietnam’s population and Vietnam’s rice production demonstrate the feasibility and applicability of the proposed methods.

Some results of classification problem by Bayesian method and application in credit operation

ABSTRACT This study proposes some results in classifying by Bayesian method. There are upper and lower bounds of the Bayes error as well as its determination in case of one dimension and multi-dimensions. Based on the proposals for estimating of probability density functions, calculating the Bayes error and determining the prior probability, we establish an algorithm to evaluate ability of customers to pay debts at banks. This algorithm has been performed by the Matlab procedure that can be applied well with real data. The proposed algorithm is tested by the real application at a bank in Viet Nam that obtains the best results in comparing with the existing approaches.

Clustering for Probability Density Functions by New k-Medoids Method

This paper proposes a novel and efficient clustering algorithm for probability density functions based on -medoids. Further, a scheme used for selecting the powerful initial medoids is suggested, which speeds up the computational time significantly. Also, a general proof for convergence of the proposed algorithm is presented. The effectiveness and feasibility of the proposed algorithm are verified and compared with various existing algorithms through both artificial and real datasets in terms of adjusted Rand index, computational time, and iteration number. The numerical results reveal an outstanding performance of the proposed algorithm as well as its potential applications in real life.

Similar Coefficient for Cluster of Probability Density Functions

ABSTRACT In this article, we propose a new criterion to evaluate the similarity of probability density functions (pdfs). We call this the criterion on similar coefficient of cluster (SCC) and use it as a tool to deal with overlap coefficients of pdfs in normal standard on [0;1]. With the support of the self-update algorithm for determining the suitable number of clusters, SCC then becomes a criterion to establish the corresponding cluster for pdfs. Moreover, some results on determination of SCC in case of two and more than two pdfs as well as relations of different SCCs and other measures are presented. The numerical examples in both synthetic data and real data are given not only to illustrate the suitability of proposed theories and algorithms but also to demonstrate the applicability and innovation of the proposed algorithm.