— 1" author: “Jintian Lu” date: “5/30/2018” output: html_document: default pdf_document: default —

The definition of Perception (from the Latin perceptio, percipio) is the capability of identifying and interpreting sensory input derived from an organism environment.

Under the data analytics context, perception in visualization can be identified as create a framework for effective data visualization of relational data that helps the data user present the large amount of data in graphical methods, interpret pattern and trend, compare different data sets and identify potential statistical features. According to a research paper on the role of visual perception in data visualization by Mehdi Dastani in 2002, an effective visualization should strongly benefit from the capabilities of the human visual system. We formulate. Therefore, the effectiveness of visualization as follows: a visualization presents the input data effectively if the intended structure of the data and the perceptual structure of the visualization coincide. In addition to the concept of the perception. The paper also provides a process model for effective data visualization. It determines that the process of perception determines perceivable relations among visual elements on the basis of their visual attribute values. As visual attribute values are assigned to visual elements by the layout process, perception is considered as the inverse of the layout process.
The importance of perception is that it is a process that needs human’s judgement based on experience or statistic models instead of determined by coding language. There are examples of people studying the role of perception in data visualization. In JOURNAL ARTICLE called Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods written by William S. Cleveland and Robert McGill, the writers of the research typically studied the science of graphs through human graphical perception . First, on how to use different charts to present data that would show the characteristics of the data sets. Essentially, the purpose of their paper was to establish and develop and test overall theories of graphical perception. They have made it clear that even though the interpretation from statistical visualization to human perception is important, it would only be defined as effective if the decoding is correct. The process of their research is as follow: First, they use hypothesize to set a series of elementary perceptual tasks. This required employ people to identify quantitative information from visualizations. The second step is set the hypothesis of levels of perceptual tasks based on the difficulty or efficiency. Then they ran the hypothesis testing to compare the theory and the reality of the results. The results are: with the Position-Angle experiment >80% of the errors were due to angular judgement (deciphering pie charts). In addition, Additionally, in the Position-Length experiment the greatest source of error (~ 25 and 45%) were due to length judgements. These results indicated that pie chart or chart with that present data with length may not be effective ways to deliver the information in the data sets.
some caption As time goes on, more researches in this area being conducted and provided us with more aspect on the relationship between the perception and data visualization with more advanced technology and a larger population. Heer and Bostock are the two researchers that trying to test visual designs and charts that affects human perceptions. They first introduced the concept of crowdsourcing. They pointed out that it is an effective way to conduct an experiment as the cost of the of recruiting is relatively low and the access to test subject is easier, especially for easier tasks with low demand for timing or higher skills. Then they specified that they conducted their research through Amazon Turk. However, the limitation to utilize Amazon Turk is that it is hard to ensure the credibility of the participants as they are unable to be observed to ensure the quality of their work. Their discovery is that crowdsourcing of graphical perception studies can be conducted. Although there are disadvantages, the variation of results could be compensated by the convenience of gathering subjects, as the researchers are able to collect more test subjects to participate at the lower cost.

Another paper studied how color would affect the data visualization. Wang, Giesen, McDonnell, Zolliker and Mueller published their paper in 2008 named ‘Color Design for Illustrative Visualization’. The purpose of the research was to assist users to select colors for scene object. They identified that there was a gap between human perception in data visualization and the color recognition. They addressed the question in a different perspective- one in which a volume or image has been divided into a set of coherent regions. It created a new framework that aimed to aid users in the selection of colors for scene objects and work cohesively with individual preferences, important functions/ formulas and overall visualization. The new framework also provided a new perspective in combining, and selecting colors to present semi- transparent layers. The paper also introduced a wide range of industrial acknowledgement or rules in using colors for their data visualization. By using a color wheel, the researchers were able to select and mix and match colors to fit into the data. some caption

The definition of Perception (from the Latin perceptio, percipio) is the capability of identifying and interpreting sensory input derived from an organism environment.

The Role of Visual Perception in Data Visualization