【摘要】：大數(shù)據(jù)時代中,每一條數(shù)據(jù)都蘊含巨大的價值,但是很少有企業(yè)意識到可以通過數(shù)據(jù)可視化將這些數(shù)據(jù)轉換為實際的經(jīng)濟價值,而這很大程度上是因為當前的許多數(shù)據(jù)可視化系統(tǒng)表達能力較弱,同時也缺乏一些智能性,使決策者無法及時作出正確的決策。本文設計并實現(xiàn)了一個數(shù)據(jù)智能可視化系統(tǒng),解決了兩個最核心的技術難題:1)具有高可擴展性的圖形透視表配置的生成;2)如何給用戶推薦更有價值的圖形透視表配置,而這又包括"下一步怎么看"和"應該怎么開始看"這兩個重要問題。針對圖形透視表配置的生成,本文提出了新的表代數(shù)算子,并用其生成透視結構配置,然后基于改進的圖形語言來生成圖形設計,并歸納了圖形透視表配置的顯式信息與隱式信息。而在智能推薦的問題上,本文總結并設計了基于數(shù)據(jù)特征的標記類型推導規(guī)則;提出了數(shù)據(jù)特征組合的三原則,并結合先驗知識設計了啟發(fā)式算法解決了單字段配置問題;設計了基于優(yōu)先原則的多字段圖表類型優(yōu)先級算法,提出了基于圖形語言的圖形透視表配置推薦算法。本文設計實現(xiàn)的系統(tǒng)已經(jīng)作為商業(yè)智能可視化平臺網(wǎng)易有數(shù)的一個模塊,應用在網(wǎng)易多款產(chǎn)品中,為運營決策提供了能夠進行數(shù)據(jù)探索的智能可視化工具。
[Abstract]:In the era of big data, every piece of data contained great value, but few enterprises realized that they could convert the data into actual economic value through data visualization. This is largely due to the weak expression ability of many current data visualization systems and the lack of some intelligence, which makes the decision-makers unable to make the correct decision in time. In this paper, a data intelligent visualization system is designed and implemented, which solves two core technical problems: 1) the generation of high expansibility graphics perspective table configuration; 2) how to recommend more valuable graphical PivotTable configuration to users, and this includes two important questions: "what to think next" and "how to start looking". In this paper, a new table algebra operator is proposed, which is used to generate the perspective structure configuration, and then the graphic design is generated based on the improved graphic language, and a new table algebra operator is proposed for the generation of the configuration of the graphic PivotTable. The explicit information and implicit information of graphic PivotTable configuration are summarized. On the issue of intelligent recommendation, this paper summarizes and designs the derivation rules of tag types based on data features, puts forward three principles of data feature combination, and designs a heuristic algorithm based on prior knowledge to solve the problem of single-field configuration. This paper designs a multi-field chart type priority algorithm based on the priority principle, and proposes a graphic PivotTable configuration recommendation algorithm based on graphics language. The system designed and implemented in this paper has been used as a module of NetEase, a business intelligence visualization platform, which has been used in many NetEase products, and provides intelligent visualization tools for operational decision-making that can carry on data exploration.
【學位授予單位】：浙江大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP311.52

【參考文獻】

相關期刊論文 前1條

1 戴國忠;陳為;洪文學;劉世霞;屈華民;袁曉如;張加萬;張康;;信息可視化和可視分析:挑戰(zhàn)與機遇——北戴河信息可視化戰(zhàn)略研討會總結報告[J];中國科學:信息科學;2013年01期

，

本文編號：2457730