interactions in time; evaluation and redesign of three abstract temporal data visualisations
TRANSCRIPT
Interactions in timeEvaluation and redesign of
three abstract temporal data visualisations
Lisa KoemanChristopher Power
Author:Supervisor:
“a non spatial continuum that is measured in terms of events which succeed one another
from past through present to future”[Merriam-Webster Dictionary, 2012]
time
“a non spatial continuum that is measured in terms of events which succeed one another
from past through present to future”[Merriam-Webster Dictionary, 2012]
time
past present future
past present future
YYYY-MM-DD, Event 1YYYY-MM-DD, Event 2YYYY-MM-DD, Event 3
etc.
temporal data
YYYY-MM-DD, Event 1YYYY-MM-DD, Event 2YYYY-MM-DD, Event 3
etc.
temporal data
past present future
YYYY-MM-DD, Event 1, ValueYYYY-MM-DD, Event 2, ValueYYYY-MM-DD, Event 3, Value
etc.
time-series data
past present future
visualisation of temporal data
• evolved over many centures
• little knowledge on the visualisations; lack of empirical evaluations
• which types of visualisations are most appropriate for which kinds of tasks?
YYYY-MM-DD, Event 1, ValueYYYY-MM-DD, Event 2, ValueYYYY-MM-DD, Event 3, Value
etc.
raw data
difficult to interpret &
time-consuming
visualisation of temporal data
• evolved over many centures
• little knowledge on the visualisations; lack of empirical evaluations
• which types of visualisations are most appropriate for which kinds of tasks?
YYYY-MM-DD, Event 1, ValueYYYY-MM-DD, Event 2, ValueYYYY-MM-DD, Event 3, Value
etc.
difficult to interpret &
time-consuming
data visualisation
raw data
visualised data
visualisation of temporal data
• evolved over many centures
• little knowledge on the visualisations; lack of empirical evaluations
• which types of visualisations are most appropriate for which kinds of tasks?
YYYY-MM-DD, Event 1, ValueYYYY-MM-DD, Event 2, ValueYYYY-MM-DD, Event 3, Value
etc.
data visualisation
“the use of computer-supported, interactive, visual representations of data to amplify cognition” [Card et al, 1999]
raw data
visualised data
digitally visualised data
difficult to interpret &
time-consuming
• evolved over many centures
• little knowledge on the visualisations; lack of empirical evaluations
• which types of visualisations are most appropriate for which kinds of tasks?
but are they any good?
method
three visualisations, three datasets& set of identical task kinds
within-participants design
1 2 3
task kindsquestion 1
question 2
question 3
question 4
question 5
question 6
question 9
existence of a data elementexample: “was a measurement made on 8 December 1977?”
temporal locationexample: “when was the lowest number of births?”
rate of changeexample: “how much is the difference in number of births between1 February 1977 and 1 February 1978?”
sequenceexample: “did the number of births reach 331 before or after Marchin 1982?”
question 7
question 8
temporal patternexample: “when you look at the overall visualisation, do you see anypatterns in the data?”
[MacEachren, 2004]
visualisation 1: calendar
[M. Bostock, On-line]
visualisation 2: timeline
[Shutterstock, On-line]
visualisation 3: radial
[Tominski and Hadlak, On-line]
measurements
completion time accuracy of answers
perceived ease of use preference
✓x
measurements
... and qualitative data on positive & negative aspects of each visualisation -
and suggestions for improvement
+ observations
+ -
participants
18 participants (1 female, 17 male)
all part of Computer Science department
mean age of 26.2 years (ranging from 20 to 36)
results: completion time
0
25
50
75
rate of change sequence
calendar visualisationtimeline visualisationspiral visualisation
existence of data
element task
temporallocation
significantly shorter completion time in calendar visualisation
seco
nds
results: accuracy
existence of data
element task
temporallocation
accuracy is significantly higher in timeline visualisation, compared
to calendar visualisation
0
25
50
75
100
rate of change sequence
calendar visualisationtimeline visualisationspiral visualisation
perc
ent
results: ease of use
calendar visualisation was perceived as significantly easier
to use than the spiral visualisation
0
2
5
7
9
easy to use difficult to use
calendar visualisationtimeline visualisationspiral visualisationfr
eque
ncy
very easyto use
neither easynor difficult
very difficultto use
results: preference
preferences are significantly different froman even distribution: timeline visualisationis preferred by the majority of participants
calendar
timeline
spiral
no preference
0% 15% 30% 45% 60%
11,11%
5,56%
55,56%
27,78%
percent of participants who preferred this option
comments
content analysis on positive aspects, negative aspects and suggestions for improvement:
kappa coefficient of 0.91
using the qualitative feedback, redesigns of all visualisations were produced
taskpresentationneither
explanations: calendar
[M. Bostock, On-line]
redesign: calendarSundayMondayTuesdayWednesdayThursdayFridaySaturday
1902
SundayMondayTuesdayWednesdayThursdayFridaySaturday
1903
SundayMondayTuesdayWednesdayThursdayFridaySaturday
1904
SundayMondayTuesdayWednesdayThursdayFridaySaturday
1905
January February March April May June July August September October November December
January February March April May June July August September October November December
January February March April May June July August September October November December
January February March April May June July August September October November December
Show dates 0 - 20%
21 - 40%
41 - 60%
61 - 80%
81 - 100%
Edit ranges...
=
=
visualisation 2: timeline
[Shutterstock, On-line]
redesign 2: timeline
50
100
150
200
250
300
350
1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912
Date: 02-12-1909 Value: 160
1900 19101980 1920
Start date: 01/01/1902 End date: 03/11/1912
visualisation 3: radial
[Tominski and Hadlak, On-line]
redesign 3: radialRange: 1994 - 1998
Zoom: + -Navigate to: dd/mm/yyyy
0 - 20%
21 - 40%
41 - 60%
61 - 80%
81 - 100%
Edit ranges...
Jan
Feb
Mar
Apr
May
JunJul
Aug
Sep
Oct
Nov
Dec
1994
1995
1996
1997
1998
Preview of zoom:
conclusions• significant differences found in task kinds
carried out in calendar, timeline and radial visualisation: completion time, accuracy, perceived ease of use and preference
• preference differs from actual measured “performance” of participants, as does familiarity
• informal evaluation of redesigns shows improvements can be made
• results show that empirical evaluations give insights that have implications for design
limitations of study
• debatable: evaluating data visualisations using pre-defined tasks
• three specific implementations of types of visualisations
• different levels of familiarity with visualisations
• ideally, exact same tasks should be compared, in exact same datasets
• participants not representative
future work• more empirical evaluations of data
visualisations: better understanding of components that influence performance
• ensures quicker, more accurate performance, essential for many professional domains
• working visualisations of redesigns should be evaluated in similar fashion
• developing evaluation method that covers real life interaction with visualisations
• what users want vs. what is best for them
references
• Merriam-Webster Dictionary, “Definition of ‘time’,” [On-line]. Available: http://www.merriam-webster.com/dictionary/time.
• S. Card, J. Mackinlay, and B. Shneiderman, Readings in information visualization: using vision to think. Morgan Kaufmann, 1999.
• A. MacEachren, How maps work: representation, visualization, and design. The Guilford Press, 2004.
• M. Bostock, “Calendar visualisation with D3.js,” [On-line]. Available: http://d3js.org/.
• Shutterstock, “Rickshaw visualisation,” [On-line]. Available: http://code.shutterstock.com/rickshaw/.
• C. Tominski and S. Hadlak, “Spiral visualisation,” [On-line]. Available: www.informatik.uni-rostock.de/~ct/software/TTS/TTS.html, University of Rostock.