32146 Data Visualisation and Visual Analytics : Graph Visualization

Watson Analytics is a graph visualization and smart data analysis service that can be used to rapidly determine meaning and patterns and in our data. It also provides guided data discovery, cognitive capabilities such as natural language dialogue and automated predictive analytics, so that interact with data is done for getting the answers.

A business intelligence tools do not naturally support inquiring or visualizing network data. The data that is to be represented in the graph is obtained, the flight data with six attributes is provide. The given data has the data regarding aircraft class, from and to city, aircraft design, engine design along with the price. Since the attributes are given in separate column, the visual route map is provided for each airline by associating it with its origin and the destination city. This can be easily done by loading up the information in Watson Analytics and a new data exploration quickly started.

With the assistance of Watson Analytics the connections between each origin and destination city in the data can be made. By clicking the maximum related proposal it shows all connections between. It also detects destination state forms a hierarchy with destination city and it capacity to auto included state names for each city. Each city is gives as node and the line is given as the edge, the line is the flight connecting the two cities (Badam and Elmqvist, 2017).

Answer 2

 A sequence of phases that is passed out to abstract information from the raw data is defined as visual data processing cycle. Each phase essential to be in order and this order is to be in cyclic. The storage stage and output results in repetition of the data collection stage that results visual data processing in another cycle of the.

1. Information collection:The first stage in the visual data processing this is very critical because the superiority of the data that is collected have tendency to impact the output heavily. It is essential to ensure the data that is gathered is both accurate and well defined. This phase provides the baseline to measure the target and what is to be improved. The data collection includes the statistical population or data collection in group. During the data collection a sample survey is done only on the selected sample (Börner et al., 2015).

2. Preparation of data:Data preparation is the manipulation of data into the procedure that is appropriate for advance visual data processing and analysis. Unprocessed data must be check for accuracy because processing cannot be done further. Constructing a dataset from single or more number of sources of data that is to be used for additional processing exploration is termed as preparation. Highly misleading results that are heavily dependent on the quality of data prepared can occur if examining of data that has not been sensibly screened for problems that can be produce further (Torre, 2014).

3. Input: It is the process where the data is confirmed and converted otherwise it can be coded into a machine readable form that a computer can able to process it. With the help of a digitizer, keyboard, scanner, or data entry from an existing source data entry is done. Speed and accuracy is required for this time consuming process. A formal and a strict syntax is to be followed then a great transaction of dispensation of power is required to break the composite data at this phase. Many productions are resorting to outsource this phase as it cost more (Ghani, Elmqvist and Yi, 2012).

4. Visual Data Processing:When the data is exposed to numerous methods and means of manipulation, and the point where a program is to be executed in a computer, and it also contains the program code. Depending on the operating system, the data process is made up of multiple threads of execution that concurrently execute instructions. While a computer program is a passive collection of instructions, a process is the actual execution of those instructions. For processing large volumes of data within a shorter period many software are used (Biedl and Pennarun, 2017).

5. Interpretation and Output: The information that is processed is further transmitted to the user in this stage. Here the output is presented as the graph visualization. Output is needed to be construed because it can provide a significant information that will be monitored for forthcoming decisions (Brandes and Pich, 2011).

6. Storage: The final phase in the visual data processing cycle is the data and the information is store for future usage. It allows quick access and retrieval of the processed data, permitting data to be passed on to the subsequent stage directly whenever needed, these are the key features. For holding system and application software ever computer uses storage (Kupczok, Schmidt and von Haeseler, 2010).

Answer 3

1. Layout Design Specification
2. Edge-crossing problem

For dealing with the Edge-crossing problem, each node in the graph can be expanded by pushing or bending the edges out. By looking into the straight equilibrium like molecular structure and by using splines while implementing a boundary on proximity. The whole diagram can be scaled if the desired shape is not obtained (Zakharov and Zhiznyakov, 2015).

1. Node-overlap problem

For removing the node-overlap spring algorithm can be used. The present spring algorithms is uses a static natural length called k for calculating the spring forces. The structure that is framed usually will be a decent drawing for a graph visualization. There may be some overlapping nodes when the labels are added (Niu, Yang and Deng, 2013). To fix this overlapping problem variants of the spring embedder model called Orthogonal Dynamic Natural Length Spring (ODNLS) and Dynamic Natural Length Spring (DNLS) can be used (Madhusudanan Pillai, Hunagund and Krishnan, 2011).

1. To enhance the readability of the layout

For enhancing the readability of the layout the following scaling algorithm can be used. The view of centrality is to adopt on suitable edge lengths in the layout. The concept centrality is often used in social network analysis, whereby a combinatorial function that is applied to each and every vertex in the network to quantify that vertex’s importance that is relative to other vertices can be made (Lee, 2015). 

1. Labelling techniques

The labelling techniques are given as follows

• Hierarchical Drawing Methods
• Dynamic graph drawing
• Planar Layout Techniques
• Optimizing Aesthetic Criteria
• Force Directed Layout Techniques

1. Graphic Design Specification
   1. Graphic objects design

Graphic object design, is the skill and the practice of development and projecting experiences and ideas with the visual and written content. The form of the communication can be virtual or physical, and it may include graphic forms, images and words.

1. Graphic attributes design

These are the attributes

• Connectivity
• Depth
• Access frequency
  1. Mapping domain-specific attributes to graphic attributes

The domain-specific attributes can be mapped by following methods

• The Geometric Mapping
• The abstract graph representation
• The Graphical Mapping
  1. To address data scale problem

The data scale problems can be address with the following

• Mislabeling
• Not to Scale
• Impossible Comparisons
• Valuing Form over Substance
  1. To enhance the readability of domain-specific attributes

The domain-specific attributes can be enhanced by graphical attributes associated with glyphs and line types and creating the pictorial representation. It also can be enhance by using the color effectively.

Answer 4

1. Transformations specification
2. Views design and transformation algorithm

Graph can be visualized with the help og a layout and also with graphical representation of its adjacent matrix. For the study of the transformation of the graph, node-link layout can be used. The graph visualization is more important.

A basic trigonometric relation where the length of the chord x is equal to

 x = t ∗ sin(θ)/θ

Then Euclidean distance can be calculated between the points by the formula

(2)

For minimization of the sum of arcs the Lombardi property is kept

From 2 and 3 the above equation is obtained

The equally spaced tangents of the Lombardi property the arc edges remains preserved whereas by moving the central vertex on the plane. By moving central vertex Pd i=1 → Fi= 0 may be obtained

 → Fi mentions to the spring force that is exerted between the central vertex and i-th neighbor vertex by assuming zero that rest length on each spring.

A Barycentric methods is applied to G= (V, E)

The vertices of the graph according to descending order of degree is arranged;

Let v = |V | and vi be the ordering

There will be no tangents in the beginning, they are formed in the FOR

for i = 1 to v do

for j = 1 to d-1 do

j ← j + 1

end for

end for

1. Animated viewing algorithm

A viable node to become the new focus is selected for exploring the graph. To specify how the visualization is connected or applied to the algorithm this is an important practical task in creating algorithm visualizations. It can be classifies as event driven, state driven, automatic animation and visual programing (Mao and Li, 2012).

1. Human cognition process

Human memory is measured as information processing system. Sensory memory, short-term memory and long-term memory these are the three basic components. The information that stored in the long-term memory is also recovered back to the working memory for processing once it is needed.

For emphasizing the limitations of working memory cognitive load theory is used (Bhowmick et al., 2014). 

Visualization efficiency is to be considered for the three-dimensional method of visualization measuring.

1. HCI
2. Fitts's law

For example, proceed mouse to specific position in the UI application and specify it for target acquisition activity. The fitts's law explains the way in which the distance to the goal from the start point and target width determines the ID index activity (Li, 2008).

Here,

A - Activities amplitude or distance

W - Width of target

ID - Index of Difficulty

Test data evaluation

The below plots illustrates the test data evaluation. It automatically update after two seconds. IDe is calculated as

Here, De - mean distinct between start and end point

We - effective width

We is 4.133 times σ

Here,

σ – Hit point Standard deviation on target

Calculating σ in the direction of target and perpendicular to the target direction and use smaller-of heuristic.

Scatter plot of time over Ide

Throughput histogram for ever set of data

Throughput is calculates as the ration of effective index of difficulty to the movement time and the unit is bits per second. The performance of device indicators are through put and through put distribution. Matrices are used to produce encompassing assessment. For example user satisfaction, error rate and overall user comfort (Patterson et al., 2014).

The above figure shows the speed and path movement of mouse while conducting the test. The entire data’s is connected in the beginning of every trial. For making comparison across the approach direction the data movement position is projected in divergence from the linear path over distance path in px.

References

Badam, S. and Elmqvist, N. (2017). Visfer: Camera-based visual data transfer for cross-device visualization. Information Visualization, p.147387161772590.

Bhowmick, S., Dyreson, C., Jensen, C., Lee, M., Muliantara, A. and Thalheim, B. (2014). Database Systems for Advanced Applications. Cham (Alemania): Springer.

Biedl, T. and Pennarun, C. (2017). Non-aligned Drawings of Planar Graphs. Journal of Graph Algorithms and Applications, 21(5), pp.915-937.

Börner, K., Maltese, A., Balliet, R. and Heimlich, J. (2015). Investigating aspects of data visualization literacy using 20 information visualizations and 273 science museum visitors. Information Visualization, 15(3), pp.198-213.

Brandes, U. and Pich, C. (2011). More Flexible Radial Layout. Journal of Graph Algorithms and Applications, 15(1), pp.157-173.

Ghani, S., Elmqvist, N. and Yi, J. (2012). Perception of Animated Node-Link Diagrams for Dynamic Graphs. Computer Graphics Forum, 31(3pt3), pp.1205-1214.

Kupczok, A., Schmidt, H. and von Haeseler, A. (2010). Accuracy of phylogeny reconstruction methods combining overlapping gene data sets. Algorithms for Molecular Biology, 5(1), p.37.

Lee, H. (2016). Animated characters of Disney animation using the transformation and alter ego of fantasy. Cartoon and Animation Studies, 44, pp.117-141.

Lee, Y. (2015). Represented by the Color Image Emotion Emotional Attributes of Size, Quantification Algorithm. Cartoon and Animation Studies, 39, pp.393-412.

Li, F. (2008). Advances in web-based learning, ICWL 2008. Berlin: Springer.

Madhusudanan Pillai, V., Hunagund, I. and Krishnan, K. (2011). Design of robust layout for Dynamic Plant Layout Problems. Computers & Industrial Engineering, 61(3), pp.813-823.

Mao, E. and Li, X. (2012). New trends in mechatronics and materials engineering. Durnten-Zurich, Switzerland: Trans Tech Publications.

Niu, D., Yang, S. and Deng, M. (2013). Principles of Interaction Design Research Based on Cognitive Characteristics. Applied Mechanics and Materials, 347-350, pp.3446-3449.

Patterson, R., Blaha, L., Grinstein, G., Liggett, K., Kaveney, D., Sheldon, K., Havig, P. and Moore, J. (2014). A human cognition framework for information visualization. Computers & Graphics, 42, pp.42-58.

Torre, D. (2014). Cognitive Animation Theory: A Process-Based Reading of Animation and Human Cognition. Animation, 9(1), pp.47-64.

Zakharov, A. and Zhiznyakov, A. (2015). Synthesis of Three-Dimensional Models from Drawings Based on Spectral Graph Theory. Applied Mechanics and Materials, 756, pp.598-603.