"A difficulty with combined bivariate visualizations is that the connection between the individual displays has to be established by the observer mentally. That is, as the eyes move from one bivariate display to the next, the observer has to keep track of the visited dots in order to form a complete understanding of data tuples. Visualization techniques based on polylines aim to tackle this difficulty. The basic strategy is to create m axes, one for each attribute, and n polylines, one for each data tuple. The polyline of an m-variate data tuple is constructed as follows. For each attribute value of the data tuple, a position is computed at the corresponding attribute axis. The m positions that we obtain are then connected to form the polyline that represents the entire tuple." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"A scatter plot consists of two orthogonally aligned axes that represent the value ranges of two data variables. Dots are placed in the space spanned by the axes in order to visualize the data elements. Conceptually, this corresponds to a mapping of data to position. A first data variable is mapped with respect to the horizontal x-axis, and a second variable with respect to the vertical y-axis." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"A stream graph is a technique for visualizing multivariate temporal data with a linear arrangement of time. As in the previous two examples, time is shown along the horizontal display axis from left to right. The multivariate data attributes are visualized as stacked streams, there is one stream for each attribute. The actual visual encoding is based on varying the thickness of the streams along the horizontal axis. That is, the vertical height of a stream at a particular horizontal position represents the underlying data value at the corresponding time. Various alternatives exist for ordering the streams and shaping the overall stack of streams." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"An important property of a data domain is its scale. The scale determines what relations and operations are possible for the data values in the domain. At the top level, we can differentiate qualitative (or categorical) and quantitative (or numerical) data. At a second level, we can further categorize qualitative data into nominal and ordinal data, and quantitative data into discrete and continuous data." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"Description is all about characterizing an observation by the associated data elements, and thereby deriving a specification for an observation. For example, an outlier can be described by its characteristic values and, if available, its spatio-temporal context. A proper description may serve as a basis for configuring further analysis steps. In particular, a description allows for sharing first insights with other people, who can later be involved in verifying the analysis results." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"Explanation means identifying all contributing data and finding the main causes behind an observation. This involves investigating several questions. Is the observation by itself significant or did we just interpret too much into the noise among the data? Does the observation re-occur throughout the data or are we looking at a singular outlier produced by unli kely circumstances? If the observation does re-occur, does it show up reliably under the same conditions, thus forming a pattern, or are its appearances seemingly random?" (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"Node-link, matrix, and implicit representations are suited for different graph data. Node-link diagrams are good for sparse networks, which have a moderate number of edges. Dense networks with many edges are best visualized using a matrix. Trees, as we just said, are nicely represented by implicit approaches." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"Often, finding the spatial scale that best matches the task at hand is a trial-and-error procedure. It may even be necessary to create further spatial scales by subsuming or subdividing spatial units. Coarser scales can be derived from the original scale by means of a suitable aggregation strategy. This includes the application of aggregation functions such as average, sum, or count. For the creation of finer scales, a suitable distribution strategy is required to assign data values to the newly specified sub-regions. Usually, additional context information is necessary to arrive at semantically meaningful aggregations and distribution." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"Presentation is to communicate confirmed analysis results. While explanation and confirmation were about convincing ourselves, presentation is about convincing others of what we have found in the data. This is best done by telling a story about the data, the analysis, and the results. Such a story can act at different levels of emphasis. We may inform an audience by letting the results speak for themselves, explicate the results to an audience, or even persuade an audience into agreement with the results. The audience in this context can be the listeners of a talk, the readers of an article, or colleagues participating in a scientific discussion." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"The simple, yet very effective idea of table-based visualization is to retain the tabular layout of spreadsheets, but to replace the textual representation of data values by a visual representation. A visual representation will not only make the interpretation of the data much easier, it will also require less display space." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"The advantage of sequencing views in time is that each view can fully utilize the display space. There is no need to divide the space among views. Obviously, sequencing views in time is particularly suited to convey temporal characteristics of data. It can also be helpful to take the user on a journey from one data facet to another. However, presenting views in quick succession to the user also has some limitations. For example, it could be difficult to make sense of all the information provided during a sequence of views. Especially when sequences take a long time, users may be unable to follow and could drown in an indigestible flood of visual representations. Therefore, it is mandatory to provide interactive controls to pause, slow down, reverse, and advance the presentation." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"The cycle plot is a technique particularly designed for the combined visualization of linear and cyclic components of temporal data. The basic idea is to show the cyclic component as a line plot into which several smaller plots are embedded to visualize the linear component. As such, the cycle plot is a kind of nested visualization." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"The triangular model is a technique particularly for visualizing intervals. It is based on two coordinate axes, the horizontal one representing time and the vertical one representing duration. In the triangular model, an interval is represented as a dot with two attached arms. The dot is placed so that the arms connect the time axis exactly at the start and the end of the represented interval. The point’s height corresponds to the interval’s duration." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"The triangular model is useful when it comes to reasoning about properties and the relationships of multiple intervals, because it generates easily distinguishable visual patterns for all possible interval relations. There is even room for visualizing data that might be associated with the intervals. The dot-based encoding would allow for resizing or coloring the dots based on some attribute values. Yet, the triangular model is only of limited use for multivariate attributes." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"When the data to be analyzed become more complex, it is no longer feasible to indiscriminately present each and every aspect of the data in a single view. When we reach this point, it makes sense to create several dedicated visual representations, each focused on communicating a particular aspect or facet of the data. The question is how several such views can be presented to the user in order to convey a comprehensive picture?" (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
"With each variable being added to the visual mapping, the richness of the visual representation is increased. Theoretically, we could add yet another visual variable, for example, by texturing the shapes. However, from a practical point of view, there are limits. While a rich visual mapping opens up the possibility to make a wider range of analytic discoveries, the downside is that the mental effort required to digest the visual representation increases as well. Therefore, it is really important to balance the visual mapping according to the task and the data." (Christian Tominski & Heidrun Schumann, "Interactive Visual Data Analysis", 2019)
No comments:
Post a Comment