The research in market segmentation includes two main parts. We first focus on discussing the market segmentation problem by applying clustering technique in data mining discipline. The partition of market is based on...
详细信息
ISBN:
(纸本)0780384032
The research in market segmentation includes two main parts. We first focus on discussing the market segmentation problem by applying clustering technique in data mining discipline. The partition of market is based on users' preference data and not on the commonly used one, users' attribute data. In that way, the definition of distance between two customers by their preference to a set of specified competing products is given. In stead of starting from scratch, the self-organization feature map is adopted as a basic clustering framework. In order to process the preference data, some necessary modifications are made. Both theoretic analysis and practical experiment are presented in this paper, which make us confident of that the algorithm we proposed has excellent performance and could discover the potential clustering patterns in the complex datasets. The second part is focus on displaying market segmentation structure. We apply visualization technique to representing the market structure clearly in a two dimensional plane so that the marketers can make their market strategies easier. The two main parts are organized as an integrated approach. Such an approach includes three core steps: preference data collecting step, preference data clustering step by SOM neural networks and visualization step by ideal point model. There are three main advantages of the approach: firstly, the approach is based on well-defined mathematic models and can be supported by a series of numeral methods. Secondly, it does not have to face the tough market variable selection problem because we focus on preference data, not on evaluators' attribute data (demographic or geographic data etc.). Finally, the approach can produce multi-scale view of market segmentation results. The experiments show that the approach yields meaningful results and is comparable and complemented to the most general ones.
Quantitative information visualization such as statistical graphics is concerned with the visual representation of quantitative and categorical data for statistical analysis. With improvements in graphics display tech...
详细信息
ISBN:
(纸本)1581138849
Quantitative information visualization such as statistical graphics is concerned with the visual representation of quantitative and categorical data for statistical analysis. With improvements in graphics display technology, it is now possible to make use of motion to stimulate recognition of patterns and structure embedded in quantitative data. Past studies have shown that the judged final position of a moving target is often displaced in the direction of the anticipated future motion of the target. Termed as representational momentum, these memory distortions have a strong relationship with the target's velocity. This study investigated human performance in visualizing dynamic quantitative information in augmented reality environments. Statistical results showed that the differences in speed and percentage change affect subject's accuracy in perceiving quantitative information significantly. On the other hand, the differences in display devices (Head-mounted-display and Liquid-crystal-display) did not indicate significant effects on subject's performance. Our results also showed that as the speed increases, the errors made in judging the final position of the moving bar also increases.
A new computational method to study within-host viral evolution is explored to better understand the evolution and pathogenesis of viruses. Traditional phylogenetic tree methods are better suited to study relationship...
详细信息
ISBN:
(纸本)0769521940
A new computational method to study within-host viral evolution is explored to better understand the evolution and pathogenesis of viruses. Traditional phylogenetic tree methods are better suited to study relationships between contemporaneous species, which appear as leaves of a phylogenetic tree. However, viral sequences are often sampled serially from a single host. Consequently, data may be available at the leaves as well as the internal nodes of a phylogenetic tree. Recombination may further complicate the analysis. Such relationships are not easily expressed by traditional phylogenetic methods. We propose a new algorithm, called MinPD, based on minimum pairwise distances. Our algorithm uses multiple distance matrices and correlation rules to output a MinPD tree or network. We test our algorithm using extensive simulations and apply it to a set of HIV sequence data isolated from one patient over a period of ten years. The proposed visualization of the phylogenetic tree/network further enhances the benefits of our methods.
This paper introduces a new toolbox for use with Matlab® and the Virtual Reality Toolbox 3.0. The toolbox is intended to assist in the performance analysis of multivariate systems and to help presenting the exper...
详细信息
Numerous digital atlases of the brain have been developed for different species through the efforts of researchers around the world. For a comprehensive picture of brain morphology and function, it is important to dis...
详细信息
ISBN:
(纸本)0769521959
Numerous digital atlases of the brain have been developed for different species through the efforts of researchers around the world. For a comprehensive picture of brain morphology and function, it is important to discover and bring together various images, segmentations and markup, generated within different projects, and juxtapose them within a single federated atlas interface. This paper describes strategies and tools for integrating distributed sources of brain data spatially. We focus on the organization of web-enabled spatial data sources which include ArcIMS feature and image services and distributed grid sources, and on the Smart Atlas, a GIS-based atlas environment enabling users to discover, access, visualize and query heterogeneous images and image markup. We demonstrate construction of spatial databases from unorganized vector and raster brain data based on ontological relationships between anatomical features from the Unified Medical Language System, spatial registration mechanisms' and tools and user interfaces for federated brain datavisualization and analysis.
A complete four channels acquisition system for high-resolution spectroscopy has been designed and realized. The analogue section is composed of four semi-gaussian shaping amplifiers with two software-selectable shapi...
详细信息
ISBN:
(纸本)0780387007
A complete four channels acquisition system for high-resolution spectroscopy has been designed and realized. The analogue section is composed of four semi-gaussian shaping amplifiers with two software-selectable shaping times. The shaping times (150ns and 450ns) were chosen to perform either high rate or best resolution XRF measurements with state-of-the-art multi-element Silicon Drift Detectors. The pulses amplitudes are caught by four large-bandwidth peak-stretchers whose outputs are multiplexed into a single 10 MHz 12 bit ADC. An FPGA operating at 24 MHz (or 48 MHz) clock frequency controls the whole process and stores the four spectra in the on-chip RAM, thus guaranteeing a maximum counting rate per bin of 332kcps and an overall counting rate in excess of 4Mcps. The system interfaces the host PC by means of the Enhanced Parallel Port with a custom made control software for datavisualization and analysis as in a conventional MCA system. If more than four channels have to be managed, various hoards can be operated in parallel on the same EPP bus.
3D computer modeling and visualization of geological structures is currently a topical research area. Most of common methods to be able to reconstruct 3D geological models require enough raw data and storage space. Bu...
详细信息
The proceedings contain 34 papers. The special focus in this conference is on Concept Lattices. The topics include: Preconcept algebras and generalized double Boolean algebras;a mathematical model for TOSCANA-systems;...
ISBN:
(纸本)3540210431
The proceedings contain 34 papers. The special focus in this conference is on Concept Lattices. The topics include: Preconcept algebras and generalized double Boolean algebras;a mathematical model for TOSCANA-systems;an application of logical information systems to bioinformatics;concept lattices for information visualization;browsing search results via formal concept analysis;formal concept analysis and semantic file systems;numerical analysis in conceptual systems with TOSCANAJ;congruences of finite distributive concept algebras;agreement contexts in formal concept analysis;towards a conceptual theory of indistinguishable objects;conceptual knowledge processing with formal concept analysis and ontologies;a first step towards protoconcept exploration;concept extensions and weak clusters associated with multiway dissimilarity measures;FCA in knowledge technologies;applying formal concept analysis to description logics;machine learning and formal concept analysis;a comparative study of FCA-based supervised classification algorithms;modelling tacit knowledge via questionnaire data;predicate invention and the revision of first-order concept lattices;formal concept analysis for knowledge discovery and data mining;a new incremental algorithm for constructing concept lattices;applying formal concept analysis in a data mining environment;a parallel algorithm to generate formal concepts for large data and using concept lattices for requirements reconciliation.
Depth or line profiles on polished sections only provide partial and local information about the chemical distribution at the interface between two solids. Germination and growth, which happen during the reaction phas...
详细信息
Depth or line profiles on polished sections only provide partial and local information about the chemical distribution at the interface between two solids. Germination and growth, which happen during the reaction phase, can only be explored by a three-dimensional in-situ visualization. The most powerful means presently at our disposal are: (i) for non-destructive analysis, synchrotron microtomography, with a maximum lateral resolution of 0.29 mum (ESRF, European Synchrotron Radiation Facility, Grenoble, France), and, (ii) for destructive analysis, SIMS image reconstruction (resolution 0.2 mum), with a difficult quantitative interpretation. The object of this paper is to demonstrate that a 3D analysis of a series of AES images is possible at an unrivalled spatial resolution (lateral and depth) and coupled with chemical analysis. The interaction between titanium and carbon at moderate temperatures is considered here to illustrate the method. The sample is elaborated from a polished substrate of carbon on which a thin layer of titanium is deposited by PVD. Thereafter the sample has been thermally treated under secondary vacuum. The data set comprises 50 images (squares of side 7 mum) separated by 20 run, yielding to an analysis volume corresponding to a parallelepiped of 7 x 7 x 1 mum(3). On a sample maintained at 450degreesC for 30 min, the formation of TiC islands on a thickness of about 100 run has been observed. Copyright (C) 2004 John Wiley Sons, Ltd.
Accurate estimation of vessel parameters is a prerequisite for automated visualization and analysis of healthy and diseased blood vessels. The objective of this research is to estimate the dimensions of lower extremit...
详细信息
ISBN:
(纸本)0780387880
Accurate estimation of vessel parameters is a prerequisite for automated visualization and analysis of healthy and diseased blood vessels. The objective of this research is to estimate the dimensions of lower extremity arteries, imaged by computed tomography (CT). These parameters are required to get a good quality visualization of healthy as well as diseased arteries using a visualization technique such as curved planar reformation (CPR). The vessel is modeled using an elliptical or cylindrical structure with specific dimensions, orientation and blood vessel mean density. The model separates two homogeneous regions: Its inner side represents a region of density for vessels, and its outer side a region for background. Taking into account the point spread function (PSF) of a CT scanner, a function is modeled with a Gaussian kernel, in order to smooth the vessel boundary in the model. A new strategy for vessel parameter estimation is presented. It stems from vessel model and model parameter optimization by a nonlinear optimization procedure, i.e., the Levenberg-Marquardt technique. The method provides center location, diameter and orientation of the vessel as well as blood and background mean density values. The method is tested on synthetic data and real patient data with encouraging results.
暂无评论