Browsing by Author "Mammadov, M.A."

Now showing 1 - 4 of 4

An Ensemble Technique For Multi Class Imbalanced Problem Using Probabilistic Neural Networks
(Advances and Applications in Statistics, 2018) Chandrasekara, N.V.; Tilakaratne, C.D.; Mammadov, M.A.
The class imbalanced problem is one of the major difficulties encountered by many researchers when using classification tools. Multi class problems are especially severe in this regard. The main objective of this study is to propose a suitable technique to handle multi class imbalanced problem. Probabilistic neural network (PNN) is used as the classification tool and the directional prediction of Australian, United States and Srilankan stock market indices is considered as the application. We propose an ensemble technique to handle multi class imbalanced problem that is called multi class undersampling based bagging (MCUB) technique. This is a new initiative that has not been considered in the literature to handle multi class imbalanced problem by employing PNN. The results obtained demonstrate that the proposed MCUB technique is capable of handling multi class imbalanced problem. Therefore, the PNN with the proposed ensemble technique can be used effectively in data classification. As a further study, other classification tools can be used to investigate the performance of the proposed MCUB technique in solving class imbalanced problems.
Estimating parameters of multivariate scaled t distribution of GSPC and its associated financial indices
(2015) Chandrasekara, N.V.; Mammadov, M.A.; Thilakaratne, C.D.
Identifying distributions of selected stock returns
(2015) Chandrasekara, N.V.; Thilakaratne, C.D.; Mammadov, M.A.
Modelling real world problems with Multivariate Distributions: An application to All Share Price Index and related financial indices
(Faculty of Graduate Studies, University of Kelaniya, 2015) Chandrasekara, N.V.; Mammadov, M.A.; Tilakaratne, C.D.
Modelling real world scenarios using statistical distributions became an important research area nowadays. Many real world applications associate with more than one variable. Therefore, when modelling these problems, finding the most suitable multivariate distributions reveals vast interest among scholars. This study focuses on finding the multivariate distribution of All Share Price Index (ASPI) of Colombo stock exchange and related financial indices. Findings of this study will lead to improve the accuracy of stock market prediction models and hence important for many parties in the financial sector. Analytical methods and Numerical methods can be identified in the literature which were used to find multivariate distributions. Analytical methods exhibit difficulties when the number of variables increases, as they involve heavy mathematical calculations. Numerical methods have been used by many scholars to find multivariate distributions and many were limited to combination of two or three variables. Four related financial indices can be identified with respect to ASPI as Amex Oil Index, Amex Gold Index, World Cocoa Index and GSPC index of U.S.A. stock market. Daily data from 1st August 2007 to 31st July 2012 of above mentioned financial indices were considered for the study. As all marginal distributions of aforesaid financial indices are Scaled t distributions, multivariate Scaled t distribution was considered in the study. A local optimization method with Matlab 'fmincon' function and a global optimization method with DSO algorithm were used to solve the corresponding optimization problem that involves twenty one parameters related to five dimensions of the multivariate Scaled t distribution. The results obtained for the maximum function values exhibit that the global optimization method provides substantially better estimates for the parameters of the multivariate Scaled t distribution than the local optimization method. The identified multivariate distribution of All Share Price Index and related financial indices is central, less peaked and has fat tails.