Storytelling with Data by Cole Knaflic.md 3.98 KB
Newer Older
1
2
3
4
5
6
7
# Storytelling with Data by Cole Knaflic

## Context
- exploratory vs explantory analysis
    - exploratory as discovery process
    - explanatory as communication of valuable discoveries
- identify specific audience and craft the communication and the story for **them**
8
- storyboard first (aligns with my low-fidelity sketching belief)
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

## Visuals that Work
- Preferred visualization types:
    - simple text | Text
    - table | Table
    - heatmap | Table
    - scatterplot | Graph Points
    - line | Graph Line
    - slopegraph | Graph Line
    - vertical bar | Graph Bar
    - vertical bar stacked | Graph Bar
    - horizontal bar | Graph Bar
    - horizontal bar stacked | Graph Bar
    - waterfall | Graph Bar
    - square area | Graph Area
24
- "Tables interact with our verbal system where graphs interact with our visual system, which is faster at processing information." - Knaflic
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
- Meltano could distinguish between exploratory and explanatory modes (lightweight GUI toolset in addition to potentially leveraged generated charts like avgs, extremes, etc combined with AI interestingness)
    - Auto inserted avg. dot, line separator, and color application to quickly show above and below avg. (similar to fig. 2.7)
    - summary statistic, avg, forecast point estimate, auto-outliers, other? as layering value automatically
    - we can suggest best visualization approach based on shallow and deep understanding of datasets
        - shallow as visualization type (text for detailing a number or two, line for continuous, bar as categorical, etc)
        - deep as AI interestingness or other algorithm based value extractions
- 3D graphs should be avoided due to perspective skews actually skewing data and alignment of XY tickmarks of a graph
    - My gut says iso 3D may be useful, but still be weary
- pie and donut are frowned upon, but I think her argument is weak once design, preattentive attributes, and the story telling aspects are considered
    - labeling becomes critical, but if labels don't fit cleanly, it makes these approaches subpar likely
    - these are good in my opinion for parts-of-a-whole visualization though simple text may be best

## Clutter is the enemy
- clutter negatively impacts cognitive load
- avoid diagonal lines and especially diagonal text as they increase the time it takes to read/absorb the information
- prefer left-aligned to center-aligned text where right-aligned can be more useful when anchoring an argument/message to data at the right of the graph
- "white space in visual communication is synonymous to a pause in verbal communication" - Knaflic

## Focus attention
- preattentive attributes (color, size, position) enable:
    1. direct attention
    2. hierarchy flow

## Think like a designer
- highlight important aspects (preattentive attributes in combo with preference of bold -> italic -> underline)
- eliminate distraction
- create clear hierarchy
- be intentional

## Lessons in storytelling
- people remember stories easier than data points
- setup -> conflict -> resolution
    - plot -> twists -> call to action
    - what? -> so what? -> now what?

## Recap
1. understand context
    - who are you communicating to?
    - what do they need to know?
    - what's the big idea?
    - what's the three minute story?
1. visual decisions
    - text is best for communicating a number or two
    - line charts are best for communicating continous data
    - bar charts are best for communicating categorical data (must have zero baseline)
    - avoid 3D, donut, and pie (though I disagree a bit as mentioned above)
1. eliminate clutter
    - leverage white space, alignment, and contrast intentionally
1. focus attention
    - leverage preattentive attributes like color, size, and position to focus attention intentionally
1. designer mentality
    - be intentional, logical, and aesthetically pleasing to ease cognitive load and improve speed of understanding
1. storytell
    - leverage story as stories resonate most with humans in comparison to data points alone (setup -> conflict -> resolution)