More Fisheye Views3D Graph Layout DesignsInteraction Techniques - Linking and Brushing. Making a change in one display changes other display synchronouslyFocus + Context: Bifocal Lens. General principles:- distorted view to the whole information space- focus of attention gets most space- periphery holds context information- fisheye views are examples of effective contex + focus techniques- generalizations are many. Bifocal displays:Focus + Context: Perspective Wall. Perspective Wall- details on the center panel are at least three times larger than the details on a flat wall that fits the field of view- Perspective Wall makes three times as much information possible as a flat wall that has details of the same size- smooth animation / transition of views helps users perceive object constancy- highlights relationships between objects in detail and context (objects bend around corner)- ease in adjusting the ratio of detail to context, as the user desires- intutive and easy to learn- combine with fisheye lensesFocus + Context: InfoTube. Places information into a real space:- street (similar to Motomachi street, Yokoh ama, Japan)- magazine- an “infotube” where information is placed at random (similar to large advertising on buildings like in Shinjuku, Tokyo, Japan)3D Fisheye ViewsFocus + Context: “Ryukyu Alive” Web Browser. Puts web pages into a galactic space (an information galaxy)- Ryukyu is the old name for Okinawa and means “flowing ball”- ALIVE stands for “Access Log Information Visualization Engine”- (icons of) pages recently accessed move to the outside- icons of pages with little access move to the center, get absorbed and vanish gradually- clicking on an icon will pop up the webpageDisplay of Abstracted Relationships. Most appropriately conveyed in the form of trees or graphs. Desirable features of the graph layout:- planarity (no crossing edges)- clarity in reflecting the relationships among the nodes- clean, non-convoluted design- hierarchical relationships should be drawn directional2D Graph Layout DesignsZoom and Pan. Panning- smooth movement of a viewing frame over a two-dimensional image of greater size. Zooming- increasing magnification of a decreasing fraction (or vice-versa) of a 2-D image under the constraint of a viewing frame of constant size. Transfer of the focus of attention:- zoom out --> pan --> zoom in- how to do it efficiently and while maintaining context- use space-scale diagrams. The problem: the local context is lost by the superposition of the magnified region- would like to maintain the global context while increasing the local focus (magnification)- use a fisheye lens in place of the magifying glassSemantic Zoom. Zooming affect geometric size. Semantic zooming additionally changes appearance and parts of objectsZooming While Maintaining Local Context. Assume you have a graph plotted on your screen and you would like to zoom in on a subgraph- a simple solution that is the magnifying glass (recall ghostview)Dealing with Limited Display Area. Too much data, too little display area. Must overcome limitations in screen resolution and screen space. Typical solution: scrolling. Problems with scrolling:- navigation in the whole mapped data space is difficult- large parts are hidden and abruptly switched off/on- hard to preserve a “mental map” of the entire information space. Must provide some means to maintain context- use “fisheye” scrolling techniqueBrushing in linked displays: highlighting a cluster of data in the climate-housing display automatically highlights the sane data in the longitude-latitude display.Data Exploration and Mining Techniques - The User in the Loop. View refinement and navigation loop:- view and navigation control is important for extended and detailed visual spaces that contain (visually) mapped data- working memory needs focus+context to perform betterCSE 332: Introduction to VisualizationLecture 5: Information Visualization FundamentalsKlaus MuellerStony Brook UniversityComputer Science Department(with some material taken from presentations by Profs. Jürgen Döllner, University of Potsdam, and Daniel Keim, University of Konstanz)” Klaus Mueller, Stony Brook 2003Data Analysis. Data in visualization:- digital data generated from mathematical models or computations- digital data generated from human or machine collection. Purpose of data analysis:- all data collected are linked to a specific relationship or theory- relationships are detected as patterns in the data- note: the relationship may either be functional (good) or coincidental (bad)- note: data analysis and interpretation are functionally subjective. Logical analysis:- applying logic to observations (the data) creates conclusions (Aristotle)- conclusions lead to knowledge (at this point the data become information). There are two fundamental approaches to generate conclusions:- induction- deductionInduction vs. Deduction. Induction: make observations first, then draw conclusions- organized data survey (structured analysis, visualization) of the raw data provide the basis for the interpretation process- the interpretation process will produce the knowledge that is being sought- experience of the individual scientist (the observer) is crucial- important: selection of relevant data, collection method, and analysis method- data mining is an important knowledge discovery strategy here- ubiquitious data collection, filtering, classification, and focusing is crucial. Deduction: formulate a hypothesis first, then test the hypothesis via experiment and accept/reject- data collection more targeted than in induction- only limited data mining opportunitiesThe Data. Data origin:- real world data - measured from real-world objects and processes (sensors, statistics, surveys)- model data - computed by machines (numerical simulations, scientific computations)- design data - edited by humans. Data size:- number of samples and data items (kB, GB, MB, TB). Data type:- scalar or multi-variate, N-dimensional: number of attributes per data item (attribute vector)- scalar or vector (e.g., flow direction). Data range and domain:- qualitative (non-numerical measurements) vs. quantitative (numerical measurements). Data value:- categorical (nominal): categoricies are disjunct, no intrinsic rank (e.g., {yellow, red, green})- ordinal data: data members of ordered sequence of categories (e.g. {tiny, small, large, huge})Data Exploration and Mining Techniques - The User in the