資料視覺化於生產管理報表資料

Visual Analysis of Production Data

陳紀宏

指導教授 ： 孫天龍 博士

元智大學/工程學院/工業工程與管理學系/碩士(2001年)

摘要

資料分析的工具與技術相當多，可以大概分為兩類：以電腦計算導向(computer-oriented)與以人類意識判斷導向(human-oriented)的分析。先說以電腦計算導向的工具，大致上具備的特性為：倚賴已經事先定義妥善的演算法或是電腦程式，來求得預先定義的特徵，例如統計方法、資料探勘等等，統計方法可以獲取資料的特徵或是整合性的結果，資料探勘可以依據探勘的目標如分群、分類、關連分析等等。
而以人類意識為判斷導向的分析，則是仰賴人的經驗與知識，相當有彈性辨識出特徵，例如資料視覺化將大量的資料轉換成圖形，利用人類與生俱來的視覺辨識能力，發掘資料當中隱藏的資訊，這是透過直接的數字閱讀所不容易達到的。資料視覺化尤其特別適合分析大量、動態或缺乏結構性的資料。
本研究利用形狀編碼(shape coding) 【Beddow 1996】及圖形化表格(table graph) 【郭玟琳 1999、宋保全 2000】等兩種多維度資料技術，應用於某PLC工廠以及PCB工廠的生管要點資料視覺化。本研究並透過探討視覺化的設計原則【Jones 1996】色彩的運用【何耀宗 1980, 林書堯 1981】與認知心裡學【Ware 2000】等學理，來解釋說明究竟優越的視覺能力，是如何來辨識出圖形之間的異同。

關鍵字

資料視覺化 ； 形狀編碼 ； 圖形化表格

並列摘要

Production data analysis could be roughly divided into two types: the computer-oriented and the human-oriented. Previous work on production analysis has been focused on the computer-oriented approach, i.e., using statistical, neural network, or other types of data mining algorithms to extract useful production information from the data. Seldom work stresses the “human-oriented” perspective. This research contributes to the human-oriented approach for production data analysis by developing a visualization environment for humans to see large amount of production data and extract useful information.
The motivation behind visualization-based data analysis is that by converting large amount of numerical data into a graphical representation, human users could take advantages of their natural strength in rapid visual pattern recognition to discover useful information hidden in the data that are difficult to be found by number reading. Visualization-based data analysis is especially suitable when data volume is large, or data are dynamic and lack apparent structure.
We employ two types of N-d data visualization techniques, i.e., shape coding 【Beddow 1996】 and table graph【Kuo 1999, Sung 2000】, to visualize the in a PLC manufacturing company. We also discuss the visualization design from the perspectives of design principles suggested by 【Jones 1996】, the color interactions 【Ho 1980, Lin 1981】 and the human perception【Ware 2000】.

並列關鍵字

data visualization ； Shape Coding ； Table Graph

參考文獻

Andrews, D. F., “Plots of High-Dimensional Data,” Biometrics, Vol. 29, pp. 133-139, 1972.

Beddow, J., “Shape coding of multidimensional data on a microcomputer display,” Proc. Visualization '90, San Francisco, CA, pp. 238-246, 1990.

Carriere, J., and R. Kazman, “Interacting with Huge Hierarchies: Beyond Cone Trees,” Proc. Symp. On Information Visualization, Atlanta, GA, pp. 90-96, 1995.

Chernoff, H., “The Use of Faces to Represent Point in k-Dimension Space Graphically,” Journal Amer. Statistical Association, Vol. 68, pp. 361-368, 1973.

Cleveland, W. S., The Elements of Graphing Data, Hobart Press, Summit, NJ, revised ed., 1993.