Hierarchical Data Visualanalytics Methods And Application In Project Management System

The hierarchical data from scientific research, enterprise management and social lifeis increasingly emerging. It is very practical, necessary and feasible to display the datawith intuitive view after been processed to understand that the information it contains,acquire its regularity, and aid decision-making.Hierarchical data visual analytics is the main research field of InformationVisualization and Visual Analytics, which has multiple visual method and techniques.After comparing of several common methods’ merits and defects, this article investigatedthe tree graph generating algorithm, and put forward an improved Pivot algorithmcombined with hierarchical clustering of data mining to optimize layout. Then, discussedthe dynamic interaction graphics display technology which adopts Directed AcyclicGraph and using regional dependency strategy, using Elastic Model to resolve theposition solution of transformed graph elements. This article also proposed a VisualAnalysis Model (VAM) based on Markov Logic Networks, which solved the problem thathow to get the weights of the rules and redounded to improve the accuracy ofvisualization’s result. Finally, applied hierarchical visualization methods to customercredit evaluation, quality management and document management modules of ProjectManagement System (PMS) and achieved anticipative results.This article tried various methods to improve the arrangement of Tree Map’srectangles to reduce its aspect ratio, so as to approaching square. Combining the datamining method hierarchy clustering, adopted the bottom up agglomerative method tomake those have similarity arranged near each other. So, this article presented a Pivotimprovement algorithm based on hierarchy clustering, which reached the goal ofenhancing the clarity and regularity of rectangles’ distribution.This paper introduced the concept of Markov Logic Networks to visualization andproposed VAM to determine the weighing of the relation rules between the source data.Finally the model is applied to customer credit evaluation; through the learning andreasoning historical data of evaluation indexes, obtained target data set’s indexes weights.Through experiment, the accuracy of evaluating customer credit by VAM is greater thantraditional method.