Issues of robustness and high dimensionality in cluster analysis

Finite mixture models are being increasingly used to model the distributions of a wide variety of random phenomena. While normal mixture models are often used to cluster data sets of continuous multivariate data, a more robust clustering can be obtained by considering the t mixture model-based approach. Mixtures of factor analyzers enable model-based density estimation to be undertaken for high-dimensional data where the number of observations n is very large relative to their dimension p. As the approach using the multivariate normal family of distributions is sensitive to outliers, it is more robust to adopt the multivariate t family for the component error and factor distributions. The computational aspects associated with robustness and high dimensionality in these approaches to cluster analysis are discussed and illustrated.

Roles of transmembrane domain 2 and the first intracellular loop in human noradrenaline transporter function: pharmacological and SCAM analysis

The aim was to investigate the roles of transmembrane domain 2 and the adjacent region of the first intracellular loop in determining human noradrenaline transporter (hNET) function by pharmacological and substituted-cysteine accessibility method (SCAM) analyses. It was first necessary to establish a suitable background NET for SCAM. Alanine mutants of endogenous hNET cysteines, hC86A, hC131A and hC339A, were examined and showed no marked effects on expression or function. hNET and the mutants were also resistant to methanethiosulfonate (MTS), ethylammonium (MTSEA) and MTStrimethylammonium (MTSET). Hence, wild-type hNET is an appropriate background for production of cysteine mutants for SCAM. Pharmacological investigation showed that all mutants except hT99C and hL109C showed reduced cell-surface expression, while all except hM107C showed a reduction in functional activity. The mutations did not markedly affect the apparent affinities of substrates, but apparent affinities of cocaine were decreased 7-fold for hP97C and 10-fold for hF101C and increased 12-fold for hY98C. [H-3]Nisoxetine binding affinities were decreased 13-fold for hP97C and 5-fold for hF101C. SCAM analysis revealed that only hL102C was sensitive to 1.25 mM MTSEA, and this sensitivity was protected by noradrenaline, nisoxetine and cocaine. The results suggest that this region of hNET is important for interactions with antidepressants and cocaine, but it is probably not involved in substrate translocation mechanisms.

Power system sensitivity analysis for probabilistic small signal stability assessment in a deregulated environment

Deregulations and market practices in power industry have brought great challenges to the system planning area. In particular, they introduce a variety of uncertainties to system planning. New techniques are required to cope with such uncertainties. As a promising approach, probabilistic methods are attracting more and more attentions by system planners. In small signal stability analysis, generation control parameters play an important role in determining the stability margin. The objective of this paper is to investigate power system state matrix sensitivity characteristics with respect to system parameter uncertainties with analytical and numerical approaches and to identify those parameters have great impact on system eigenvalues, therefore, the system stability properties. Those identified parameter variations need to be investigated with priority. The results can be used to help Regional Transmission Organizations (RTOs) and Independent System Operators (ISOs) perform planning studies under the open access environment.

Eigenvalue sensitivity analysis for probabilistic small signal stability assessment

This paper presents a method to analyze the first order eigenvalue sensitivity with respect to the operating parameters of a power system. The method is based on explicitly expressing the system state matrix into sub-matrices. The eigenvalue sensitivity is calculated based on the explicitly formed system state matrix. The 4th order generator model and 4th order exciter system model are used to form the system state matrix. A case study using New England 10-machine 39-bus system is provided to demonstrate the effectiveness of the proposed method. This method can be applied into large scale power system eigenvalue sensitivity with respect to operating parameters.