CONTENT
_____6.3: Visualization Graphical and Tools __________6.3.1: Commonly Used Graphics for High-Dimensional Data __________6.3.2: Visualization Software Available for Selected Graphics __________6.3.3: Initial Visualization on Selected Graphics in Tableau _______________6.3.3.1: Personal Profile & Registered User _______________6.3.3.2: Teaching Profile _______________6.3.3.3: Survey Results

6.3: Visualization Graphical and Tools

In this section, we will attempt to place the variables onto different types of visual, and select the most suitable graphic to visualize them.

6.3.1: Commonly Used Graphics for High-Dimensional Data

From the previous section, we have transformed our data such that our analysis of

Profile - will be based on categorical data display only
Survey Results/ Responses - will be based on a visual graphics of a frequency table.

We have identified commonly used graphics for Multivariate data display, and selected those that might be suitable for our analysis, as below;

Type of Graph	Profile of Respondents	Survey Results	Rationale
Trellis Display	X	X	Suitable for categorical & continuous data display
Ternary Diagram			Only suitable for 3 continuous data display
Glyphs			Only suitable for more than 3 continuous data
Parallel Coordinates			Only suitable for multiple continuous data
Heatmap		X	Suitable for tables and differentiate them by scale
Table Lens			More suitable with multiple continuous data display
Tableplots			More suitable with multiple continuous data display
Mosaic Plots	X		Suitable for categorical data display
Parallel Sets	X		Suitable for categorical data display
Tree Map	X		Suitable for categorical data display
Divergent Bar Chart		X	Suitable for categorical data display for likert scale

6.3.2: Visualization Software Available for Selected Graphics

With the selected graphic, we also highlight the relevant software that is able to produce the graphical work, as below;

Graphic/ Feature Required	Tableau	JMP	Parallel Set[Java]	Mondrian	Marcofocus-Treemap
Trellis Display	X	X
Mosaic Plots	X	X		X
Parallel Sets			X
Tree Map	X	X			X
Divergent Bar Chart	X
Heatmap	X	X
Ease of Putting into a single dashboard	X

With the above, we will be using Tableau to build our dashboard mainly, and attempt to explore the merits of parallel sets. In all, there are 3 types of information that we will like to visualize;

Personal Profile & Registered User
Teaching Profile
Survey Responses / Results

6.3.3: Initial Visualization on Selected Graphics in Tableau

6.3.3.1: Personal Profile & Registered User

The categorical variables that are included here are

Gender
Age Bin (Manual)
Registered User

The possible displays as below;

Trellis Display / Barchart	Mosaic Plots
Unable to see the % by total clearly. Thus,unable to have an overview of the % by Total.	With an overview for selection. Version 1 is too colorful and harder to relate, therefore, version 2 is preferred.

Tree Map (Version 1 - Tableau)	Tree Map (Version 2 - Macrofocus)
Difficult to visualize, and hard to differentiate the squares.	This version of the treemap is clean and simple. It is easier to visualize.

Parallel Set has been attempted, and the link as provided;

YouTube

And a snapshot of it, as below;

Note: This is currently not available in Tableau. Therefore, we are not able to consider it in the dashboard.

Conclusion: In all, Mosaic Plots (Version 2) is selected because:

It is simple and clean, with simple color scheme of 2 color only.
It is easier for human to browse from left to right (with increasing scale), instead of from up to down.
It does not take as much space, as compared to a Tree Map.
It is available in Tableau
While Tree Map produced in Macro Focus is clean and simple, but it takes up a significant space on a dashboard and the layout is not available in tableau.

Using a similar concept, we will plot the registered user as below;

Register User:A single variable , therefore; the following graphic (version 2) is selected because it is easier for human to read from left to right, as below;

6.3.3.2: Teaching Profile

The categorical variables that are included are

DOMAIN
PHD
YEARSEXP_Manual_binning_2
UNIVERSITY
UOC POSITION

In this aspect, the most important information here will be "DOMAIN", "UOC POSITION" & "YEARSEXP_Manual_binning_2".Since there are at least 3 catagorial variables that we need to visualize, we will use a duet display as below;

This is selected because it is able to display both proportional and Frequency neatly. Frequency should be display so that we are able to if representative of sub-group have sufficient sample size to draw any conclusion.

6.3.3.3: Survey Results

Last but not least, the survey results are recommended to be displayed on a Likert Scale. This option is taken with reference to this Research Paper: "Plotting Likert and Other Rating Scales", and with consultation from the Professor of the course.

The display as below;

A straight line is formed after "neutral" (0%), so as to guide the users to focus on the results that show positive of "Agree", and "Strongly Agree".