In week one, I learnt about the fundamental data types, marks and channels.
Data visualisation definition (There is no universal definition!)
definition 1
"Computer-based visualization systems provide visual representations of datasets designed to help people carry out tasks more effectively." - Munzner, Visualization Analysis & Design
definition 2
Bernie uses the framework of what, why and how to introduce them.
Data types (What)
Generally, you gotta understand three questions :
What are the types of attributes?
Categorical, 2. Ordered → (Ordinal, Quantitative)
If the data attributes can be ordered, what are the types of order direction?
What are the types of dataset?
Table (Easy to explain: cell, column and row)
2. Tree (with node and vertexes)
Spatial
Fields
Unlike table, fields stores the continuous values
Geometry
There are many position, spatially ,of which it consists of different values.
Rule : If you need more than seven colors in a chart, consider using another chart type or to group categories together.
Rule : Consider using the same color for the same variables
Rule : Make sure to explain to readers what your colors encode
Rule : Consider the color grey as the most important color in Data Visualisation
Rule : Make sure your contrasts are high enough
Rule : Consider where your colors appear in relation to each other.
Rule : Use intuitive colors
Rule : Use light colors for low values and dark colors for high values
Rule : Don’t use a gradient color palette for categories and the other way round
Rule : Use lightness to build gradients, not just hue
Rule : Consider using two hues for a gradient, not just one
Rule : Consider using diverging color gradients.
Rule : Consider color-blind people
Gestalt Principles of Visual Perception
What is Gestalt?
It means Form or Shapes.
Why Gestalt Principles of Visual Perception
When it comes to identify which visual elements are signals (the information we want to communicate) and which might be noise (clutter), consider this principles.
There are 6 Gestalt principles to follow :
1. Proximity
物以(近)聚 : we tend to think objects that are close together as a group
2. Similarity
物以類聚 : we tend to think objects that have SIMILAR color, shape, size, orientation are perceived as a group
3. Enclosure
We tend to think of objects that are physically enclosed together as a group
Leveraging the principles to draw a visual distinction
4. Closure
We tend to perceive as a set of individual elements as a single shape when they can. For example, people tend to find this as a circle first and only after that S individual element.
By default we tend to think that a chart must have a background shading and border. But when we remove The unnecessary elements our data stands out more
5. Continuity
The principal tells that when looking at objects we tend to seek smoothest path and naturally create continuity in what we see. For example if I take the object 1 apart and pull them apart we will expect to see what’s shown next
Similarly, when applying this principle to the graph, that means when I remove the unnecessary vertical Y-axis from the graph, our data stands out more. Because the consistent white space between the label on the left and the data on the right.
6. Connection
We tend to think of objects that are physically connected as a group.
So, we leverage connection principle in a line graphs, to help our eyes to see orders in the data.
Visual Hierarchy with Figure-Ground
Visual Hierarchy :
Figure-Ground
Graphical representation in which elements are ranked according to their importance.
Important elements are graphically emphasised and less important elements are de-emphasised.
Representation :
Bold, Italic, Saturation, color text
Visual depth for accentuating one object over another
Perception : one object stands in front of another and appears to be closer to the reader.
What does Figure-Ground mean ?
Figures:
important objects, become objects of attention and standout from the background
Grounds:
things less important, the background.
Layout
There are a few rules for designing layout; it sounds simple but is very essential.
1. Reading direction
We read from top-left to bottom-right
2. Visual Centre
most important things should be placed in the center
3. Sight line
An invisible horizontal or vertical grid lines that separate the visual elements
The less number of Sight line, the more stable the layout mapped.
4. Alignment with Invisible Frame
Self-explanatory
5. Symmetry and Balance
Ensure visual elements are aligned in the Symmetry manner
What is the difference between Symmetry and Balance ?
Symmetry
a balance around a central vertical axis
Balance
All elements are placed with proper distances; which is not off-balanced.
6. Balance with white space
used to group elements
There is no need to completely fill space.
Use empty space as a design element.
Typography
Things to evaluate Typography
Is it READABLE?
Is it Aesthetically appealing?
Before diving into other concepts, a number of terms needed to be understood first :
Text = what you typed
Character = a numerical code that represents the character
Glyph = a visual symbol that represents a character
Character encoding = code matches to the Glyph
Style = Italic, Bold or so
Font = Digital files of all Glyph of geometry of font
Typeface = the set of font (often misunderstood as font)
Typography = An art of text (READABLE & AESTHETIC)
2 groups of typefaces
Sans Serif = traditional
Serif = Modern
3. Attribute types of text characteristics
Categorical or Ordered
How Many Typefaces?
• Generally use a single typeface, but vary weight, size, case, italic/regular, and colour to create a visual hierarchy / figure-ground.
• If your really need, use a maximum of two typefaces, but make sure they go well together (this is difficult to get right!).
Generally combine one serif and one sans serif type family.
Tips to improve visualisation
Alignment
Hierarchy
Important notes can be bold.
Other text and colour-coded annotations can use a normal font, and special terms can use italic style.
difference between color channelsWhich Channel share the same with Identity and Magnitude channels?
There is a channel that appears in both of the two ranked lists of channels:
position.
In the magnitude channels, we have "position on a common scale", or "position on an unaligned scale". In the identity channels, we have "spatial region".
They are actually all related to the "spatial position" of the graph, which means - spatial position is the only channel that is both a magnitude channel and an identity channel, also it is the most effective one.
How to identify the type of dataset
Your example - "For example, network can be represented as an adjacency matrix, or a field can be represented has a high-dimension table etc. ? If so how do we definitively distinguish among them?"
-----
It is not based on how we store the data or how we represent the data to distinguish the dataset types. We should consider the structure of the data.
Why not the format?
1) Most of the dataset types can be possibly stored in a table or in a rational database - e.g., a network dataset can be stored as two variables: "user_1", "user_2". Each row of data with two users means they are connected.
2) A dataset stored in a json or xml format could also be table data.
Why not the representation?
for example, the adjacency matrix you mentioned is basically a 2D heatmap. So a heatmap could possibly be used to represent a table dataset or a network dataset.
So how to differentiate the data types?
you need to first understand the data types: items, attributes, like, positions, and grids.
then based on what data types are in the dataset, you can decide the dataset type: e.g., if there is a link between the items in the dataset, then this might be a network or tree dataset; if there are geo-positions, then this might be a geometry or a field dataset.
What are Fields, spatial fields and grid
Grid is related to cells:
Suppose that we want to show the distribution of the NBA shot locations - shown below. To get this, we actually need to first divide the space into a set of cells (either square cells - grid or hexagon cells); then we can get statistics of how many players have made the shots inside each cell.
Fields and spatial fields:
they are the name of the dataset that is based on cells.
What is part-to-whole relation (in pie chart) ?
part-to-whole relation is about - these four types of ingredients belong to the pancake
