Data Visualization Final Exam Review

Visual interaction is mainly divided into five categories : panning + zooming technology, dynamic filtering technology, overview + detail technology, focus + context technology and multi-view association coordination technology. Fisheye view belongs to focus + context technology visual interaction method. Multi-scale navigation belongs to the visual interaction method of pan + zoom technology; urllib.request request module; urllib.parse is the url parsing module; urllib.error exception handling module; urllib.robotparser is the robots.txt parsing module; Initiate a request and get a response; text content visualization) Text relationship visualization: Graph-based text relationship visualization (word tree, phrase network), document relationship visualization (galaxy view, document set sampling projection), theme landform (Theme map) and other text content relationship visualizations Visualization based on graphs , such as word tree, phrase nets, matrix tree diagram (NewsMap), etc. Text multi-feature information visualization methods include : Parallel Tag Clouds (Parallel Tag Clouds), FaceAtlas, etc.; in the field of medical visualization, there are three research hotspots : image segmentation technology, real-time rendering technology and multiple data collectionReflect internal structure : pie chart, stacked chart. reflect time-varying: Column chart, line chart. Reflect ranking order : bar chart. Reflecting Correlations : Scatterplots. Reflect multiple indicators: radar chart. Continuous data (discount chart, step chart, fitting curve) discrete data ( scatter plot, column chart, stacked column chart) The core process of data visualization includes data acquisition, data processing and transformation, visual mapping and user perception Four steps. Data visualization classification: scientific visualization, information visualization, visual analytics. The three elements of color: hue, purity, lightness. The main functions of data visualization include three aspects: data recording and expression ,  data manipulation , and  data analysis . Data visualization is about finding three things in data: patterns, relationships, and anomalies. The requirements for understanding text are generally divided into lexical level,  grammatical level and  semantic level . List 6 basic charts for data visualization and illustrate their use cases with examples. Data trajectory: visually present data distribution, outliers, and mean shifts, histogram: use the height of the column to reflect the difference in data, line chart: suitable for occasions where the trend is more important than a single data point, histogram: present data The distribution of outliers and the state of data distribution, pie chart: presents the proportion of each component in the whole, contour map: indicates the continuous distribution and change law of data, trend chart: uses highly dense line chart expression To show the trend of data changes with a certain variable, scatter plot: suitable for three-dimensional data sets. What are some examples of the benefits of data visualization? Data visualization makes data analysis easier. Realizing data visualization is nothing more than making people more convenient, faster and more accurate in the process of data processing. Such data analysis can not only get closer to people's lives, but also meet people's actual needs in life. Make knowledge acquisition more convenient. Data collection under the modern background has good accuracy, and the use of new software technology and means not only makes it easier for people to obtain huge databases, but also mines hidden data targets. In life, the data that is often encountered generally has an identification system. This identification system is not only directional, but also textual. For example, the guide signs on the street not only allow you to obtain information more intuitively, but also solve practical needs. Data visualization is widely used in scientific research, network and business fields. Such as the real-time dynamics of the national new pneumonia epidemic under the new crown epidemic. Basic principles of data visualization 1 Data screening. 2. Intuitive mapping from data to visualization 3. View selection and interaction design 4. Aesthetic factors 5. Metaphors of visualization 6. Color and transparency Please answer if you are not sure what are the visualization methods for data? (1) Icon method. The more common methods of the graph method include error bars, box-and-whisker plots, and flow field radar plots. (2) Geometry representation. Utilizing representative geometric objects provides richer visual expression. (3) Visual element coding method. Taking visual elements as the basic carrier of uncertainty coding is the basic idea of ​​many uncertainty visualization methods. (4) Animation expression method. What is Data Visualization:The process of mapping information into visual effects. The purpose of data visualization is to visualize data so that information can be clearly and effectively conveyed. The role of data visualization in big data analysis is reflected in: moving faster, providing decision-making suggestions in a constructive manner, and understanding the connection between data. The development history of data visualization: ancient times-1600: the germination of charts, 1600-1699: physical measurement, 1700-1799: graphic symbols, 1800-1900: data graphics, 1900-1945: modern enlightenment, 1950-1974 Years: Visual Coding of Multidimensional Information, 1975-1987: Multidimensional Statistical Graphics, 1987-2004: Multi-Interaction Visualization, 2001-Present: Visual Analytics. Challenges faced by data visualization: large data scale, data quality problems, fast and dynamic data changes, insufficient analysis capabilities, and multi-source data with different types and structures. Data visualization development direction: visualization technology is closely related to data mining; visualization technology is closely related to human-computer interaction; visualization is closely related to large-scale, high-latitude, and unstructured data. Gestalt principles: the unified perception of the whole, the principle of proximity, the principle of similarity, the principle of closure, the principle of continuity. The definition of visual encoding (visual encoding) can be summarized in one sentence: describe the mapping relationship between data and visualization results. We think of visualization as a set of graphical symbols that carry encoded information. And when people read the corresponding information from these symbols, it is called decoding. Visual coding consists of two parts: markup (graphic elements) and visual channels. Types of visual channels:Qualitative or categorical visual channels: suitable for encoding categorical data information, such as shape, color hue, spatial location. Quantitative or sequenced visual channel: suitable for encoding ordered or continuous data information, such as the length of a straight line, the area of ​​an area, the volume of a space, slope, angle, color saturation and brightness, etc. Grouped visual channel: A group is described by a combination of multiple or multiple tags. Grouping channels include proximity, similarity, and inclusion. The grouping channel is suitable for grouping data of related categories to express the inherent correlation of the data. The expressiveness and effectiveness of the visual channel : accuracy, whether the judgment results after people's visual perception are consistent with the original data. Distinguishability, the visual channel has different value ranges, how to set the value can make it easy for people to distinguish between two or more value states of the visual channel. Separability, the coded objects of different visual channels are placed together, whether it is easy to distinguish. Visual prominence, whether to encode important information with a more prominent visual channel. Big data application development process: data acquisition data preprocessing, big data storage and management, big data analysis data mining, visual display. When designing data visualization, we should abide by the following visualization design standards: 1. It must have a strong expressive power and be able to truly and comprehensively reflect the content of the data. 2. Strong effectiveness. An effective visualization can display data information in a way that is easy for users to understand in a short time. 3. Can convey information concisely, so that more data can be expressed in a limited screen, and it is not easy for users to misunderstand. 4. Ease of use, the way of user interaction should be simple and clear. It is more convenient for users to operate. 5. A sense of beauty, visual beauty can make it easier for users to understand the content to be expressed in the visualization, and improve the efficiency of the visualization View interaction includes: scrolling and zooming of the view, control of color mapping, control of data mapping methods, and data zoom tools , details control aesthetic factors: simplicity principle, balance principle, focus principle The design framework of data visualization is divided into four layers: real problem layer, abstract layer, coding layer, and interactive algorithm layer.The steps of data visualization : determine the mapping of data to markers and visual channels, and determine what data is to be presented; the selection of attempts and the design of user interaction control, the establishment of data indicators, and the gradual display of data results from overall to partial; effective screening of data , the visualized results need to maintain a reasonable information density. Difficulties in high-dimensional multivariate data analysis (1) For high-dimensional and multivariate complex data, the capabilities of visualization systems based on basic analysis and statistics are far from enough. (2) The complexity of data has greatly increased, including unstructured data and heterogeneous data collected and integrated from multiple data sources. The traditional single visualization method cannot support the analysis of such complex data. (3) The large scale of data exceeds the limit of the processing capabilities of stand-alone machines, external storage models, and even small computing clusters. The scale of data that can be processed is about GB level. New ideas are needed to solve large-scale adjustments. (4) In data acquisition and processing, data quality problems will inevitably arise, and the uncertainty of the data needs special attention. (5) Data changes rapidly and dynamically, and often exists in the form of streaming data. Real-time analysis and visualization of streaming data is still an urgent problem to be solved. (6) The traditional single visualization method cannot support the analysis of complex data such as unstructured data and heterogeneous data. (7) In the process of data collection and analysis, there will inevitably be problems such as data quality and uncertainty