【摘要】：數(shù)據(jù)是人類記錄信息的重要形式,而可視化是一種以圖形符號(hào)等更加直觀形象的方式來傳達(dá)信息的技術(shù)�？梢暬谷祟惈@取知識(shí)變得更加高效,它是人類獲取信息的重要渠道。隨著信息時(shí)代的到來,數(shù)據(jù)爆炸式增長(zhǎng),數(shù)據(jù)變得越來越復(fù)雜,數(shù)據(jù)維度較高。如何將高維數(shù)據(jù)可視化并反映數(shù)據(jù)特征和規(guī)律是當(dāng)今可視化領(lǐng)域的難點(diǎn)和熱點(diǎn)問題。本文著眼于利用可視化技術(shù)將高維數(shù)據(jù)可視化,幫助用戶發(fā)現(xiàn)數(shù)據(jù)之間的關(guān)系,數(shù)據(jù)與維度之間的關(guān)系。本文的主要研究工作如下:(1)提出了一種高維數(shù)據(jù)可視化方法。由于維度爆炸及可視空間有限,用戶很難可視化并探索、分析高維數(shù)據(jù)。早期的一些工作通過傳統(tǒng)的降維方法產(chǎn)生隱式維度,不但損失了一部分信息,更重要的是這些隱式維度很難為用戶所理解。因此,本文提出一種高維數(shù)據(jù)可視化方法,該方法結(jié)合用戶有限的知識(shí)導(dǎo)出符合用戶知識(shí)的維度,并重新組織數(shù)據(jù)。然后,利用本文基于散點(diǎn)圖矩陣擴(kuò)展的可視化呈現(xiàn)方法散點(diǎn)餅圖矩陣來可視并探索重新組織后的數(shù)據(jù)。該方法可使用戶發(fā)現(xiàn)已知數(shù)據(jù)與未知數(shù)據(jù)的關(guān)系,未知數(shù)據(jù)與導(dǎo)出維度的關(guān)系。實(shí)驗(yàn)驗(yàn)證了該方法的有效性。(2)設(shè)計(jì)并實(shí)現(xiàn)了一個(gè)高維數(shù)據(jù)可視化工具。本文通過對(duì)現(xiàn)有的可視化工具調(diào)研分析發(fā)現(xiàn),目前存在較少的高維數(shù)據(jù)可視化工具,而且現(xiàn)有的高維數(shù)據(jù)可視化工具用戶探索流程不夠完善,不易擴(kuò)展新的高維數(shù)據(jù)可視化方法。因此,亟需一個(gè)實(shí)用的高維數(shù)據(jù)可視化工具,幫助用戶更好地探索、分析高維數(shù)據(jù)。本文設(shè)計(jì)并實(shí)現(xiàn)了一個(gè)高維數(shù)據(jù)可視化工具,該工具提供一個(gè)完整的用戶可視探索數(shù)據(jù)的流程,用戶可結(jié)合交互,完成對(duì)數(shù)據(jù)的探索,并可保存數(shù)據(jù)探索結(jié)果,供用戶分享、查閱。而且,用戶可以基于該工具,針對(duì)特定的應(yīng)用擴(kuò)展新的高維數(shù)據(jù)可視化方法。
[Abstract]:Data is an important form of human record information, and visualization is a technique to convey information in a more visual way such as graphic symbols. Visualization makes the acquisition of knowledge more efficient, and it is an important channel for human to obtain information. With the arrival of the information age, data explosive growth, data become more and more complex, data dimension is higher. How to visualize the high-dimensional data and reflect the characteristics and laws of the data is a difficult and hot issue in the field of visualization nowadays. This paper focuses on using visualization technology to visualize high-dimensional data to help users discover the relationship between data and dimension. The main work of this paper is as follows: (1) A high dimensional data visualization method is proposed. Due to dimensional explosion and limited visual space, it is difficult for users to visualize and explore high dimensional data. Some of the earlier work generated implicit dimensions through traditional dimensionality reduction methods, which not only lost some information, but also were difficult for users to understand. Therefore, this paper proposes a high-dimensional data visualization method, which combines the limited knowledge of the user to derive the dimension that conforms to the user's knowledge, and reorganizes the data. Then, the scattered pie chart matrix is used to visualize and explore the reorganized data by using the visual representation method based on the expansion of scatter plot matrix in this paper. This method enables users to discover the relationship between the known data and the unknown data, and the relationship between the unknown data and the derived dimension. Experiments show that the method is effective. (2) A high dimensional data visualization tool is designed and implemented. Through the investigation and analysis of the existing visualization tools, it is found that there are few high-dimensional data visualization tools, and the existing high-dimensional data visualization tools user exploration process is not perfect. It is difficult to extend the new high dimensional data visualization method. Therefore, a practical high-dimensional data visualization tool is urgently needed to help users explore and analyze high-dimensional data better. In this paper, a high dimensional data visualization tool is designed and implemented. The tool provides a complete flow of user visual exploration data. The user can combine interaction, complete the data exploration, and save the data exploration results for users to share. Access. Furthermore, based on the tool, users can extend a new high-dimensional data visualization method for specific applications.
【學(xué)位授予單位】：北京郵電大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2016
【分類號(hào)】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 李澤安;;淺談高維數(shù)據(jù)挖掘的現(xiàn)狀與方法[J];福建電腦;2014年07期

2 陳紅彬;;基于大數(shù)據(jù)的高維數(shù)據(jù)挖掘探究[J];通訊世界;2018年03期

3 于君;范文彬;杜永軍;;智能電網(wǎng)中高維數(shù)據(jù)聚類方法研究[J];智能計(jì)算機(jī)與應(yīng)用;2016年01期

4 張凌潔;;淺談高維數(shù)據(jù)變量選擇現(xiàn)狀與方法[J];數(shù)碼世界;2016年07期

5 賀玲;蔡益朝;楊征;;高維數(shù)據(jù)空間的一種網(wǎng)格劃分方法[J];計(jì)算機(jī)工程與應(yīng)用;2011年05期

6 宋懷波;何東健;;面向精細(xì)農(nóng)業(yè)的高維數(shù)據(jù)本征維數(shù)估計(jì)方法研究進(jìn)展[J];中國(guó)科學(xué):信息科學(xué);2010年S1期

7 張航;PP型擬合優(yōu)度檢驗(yàn)[J];系統(tǒng)科學(xué)與數(shù)學(xué);1988年03期

8 余立蘋;李云飛;朱世行;;基于高維數(shù)據(jù)流的異常檢測(cè)算法[J];計(jì)算機(jī)工程;2018年01期

9 虞翔;李青;;大數(shù)據(jù)環(huán)境下的高維數(shù)據(jù)挖掘在入侵檢測(cè)中的有效應(yīng)用[J];電腦編程技巧與維護(hù);2016年22期

10 宋玉林;訾雪e，

本文編號(hào)：2252612