Samuel Huron
@cybunk
Inria - IRI

Romain Vuillemot
@romsson
Inria (now at Harvard)

Jean-Daniel Fekete
@jdfaviz
Inria
1!
Data streams are everywhere!

Mail	
  

RSS	
  feed	
  

Status	
  update	
  

Logs	
  

2
Streaming data visualization challenges:

1.	
  How	
  to	
  represent	
  incoming	
  data?	
  
2.	
  How	
  to	
  manage	
  unpredictable	
  update	
  rate?	
  	
  	
  	
  
3.	
  How	
  to	
  provide	
  context	
  +	
  details	
  over	
  Fme?	
  
4.	
  How	
  to	
  design	
  visualizaFons	
  for	
  lay	
  audiences?	
  
3
4
Metaphor: 


Physical	
   Mapping	
   Visual	
  

Data	
  Mapping	
  



from

physical 

sedimentation


Token	
  

Newest	
  
tweet	
  



to

visual 

sedimentation

!

Deposit	
  
of	
  tokens	
  

flocculaFon	
  
Aggregated	
  
Area	
  

Previous	
  
tweets	
  
Older	
  	
  
tweets	
  
5
We have
used the
metaphor
in real
world
situations

6
We have
used the
metaphor
in real
world
situations

7
We have
used the
metaphor
in real
world
situations
Average	
  visit	
  duraFon	
  4:21	
  
minutes	
  
	
  
Shared	
  1800	
  Fmes	
  on	
  
twi4er	
  
	
  
	
  

Doubled	
  the	
  number	
  of	
  
tweets.	
  
	
  
Good	
  feedbacks	
  from	
  users	
  
8
We apply this to classical charts:

9
And we have
generated a
lot of “crazy”
charts

10
How it’s done

1.	
  Design	
  	
  
	
  	
  	
  	
  	
  parameters	
  

2.	
  Toolkit	
  and	
  
	
  	
  	
  	
  parameter	
  space	
  

3.	
  ExploraFon	
  of	
  the	
  	
  
	
  	
  	
  	
  design	
  space	
  
11
The design parameters

12
Free space | Boundaries | Entry points | Forces, Strata

13
Free space | Boundaries | Entry points | Forces | Strata

14
Free space | Boundaries | Entry points | Forces | Strata

15
Free space, Boundaries, Entry points, Forces, Strata

Decay	
  
FloccullaFon	
  

Gravity	
  
16
Free space, Boundaries, Entry points, Forces, Strata

17
Layout: Templates for classic charts

18
Visual Sedimentation design inspiration

19
How to explore
a metaphor-based design space?!
In	
  this	
  paper,	
  we..	
  	
  
Create	
  a	
  toolkit	
  	
  and	
  generate	
  cases	
  studies	
  
Defined	
  a	
  parameter	
  space	
  
Generate	
  a	
  variety	
  of	
  different	
  designs	
  
Select	
  “Best”	
  samples	
  
And	
  iterate	
  	
  

20
Toolkit

Open	
  source	
  	
  
on	
  Github,	
  
GPL	
  like	
  license	
  

JavaScript	
  	
  
+	
  html	
  5	
  	
  
web	
  compliant	
  

Box2dWeb	
  is	
  
used	
  for	
  the	
  
physics	
  engine	
  

D3.js	
  is	
  used	
  for	
  
the	
  aggregated	
  
area	
  

21
Creating a chart

Apply	
  it	
  to	
  a	
  DOM	
  element	
  

var	
  mySedng	
  =	
  {	
  
	
  	
  
	
  
	
  
	
  width:	
  	
  	
  300,	
  height:	
  300,	
  
Canvas	
  :	
  
	
  	
  
	
  
	
  
	
  chart	
   	
  	
  	
  {	
  type:'StackedAreaChart',	
  …	
  },	
  
chart:	
  
	
  	
  
	
  
	
  
	
  data	
   	
  	
  	
  	
  {	
  model:	
  [	
  {label:	
  'Column	
  A'},	
  …	
  ],	
  
data:	
  
	
  	
  
	
  
	
  
	
  strata	
   	
  	
  [	
  [	
  {initValue:	
  100,	
  label:	
  'Bar	
  A'}	
  ],	
  …	
  ],	
  
strata:	
  
	
  	
  
	
  
	
  
	
  stream	
   {	
  
stream:	
  
	
  	
  	
  
	
  
	
  
	
  provider:	
  'generator',	
  refresh:	
  	
  10000/4}	
  
	
  	
  
	
  
	
  
	
  },	
  	
  
	
  
	
  
	
  
	
  sedimenta-on	
   {	
  
sedimentaFon:	
  
	
  	
  	
  	
   	
  
	
  
	
  
	
  token:	
  {	
  size:	
  {	
  original:	
  6,	
  minimum:	
  2	
  }},	
  
	
  	
  	
  
	
  
	
  
	
  
	
  aggregaFon:	
  {	
  type:'stacked’	
  },	
  
	
  	
  	
  
	
  
	
  
	
  
	
  suspension:	
  	
  {	
  decay:	
  {power:	
  1.01	
  }},	
  
	
  	
  	
  
	
  
	
  
	
  
	
  flocculaFon:	
  {...}	
  
	
  	
  
	
  
	
  
	
  },	
  	
  
	
  };	
  
	
  	
  
var	
  myChart	
  =	
  	
  $("#myDivChartContainer”).vs(mySedng);	
  
22
Toolkit parameter space

?!
23
Let’s play with this parameter space!!
and explore the design space!

24
Incoming point

25
Token visual encoding

Tokens	
  are	
  edits,	
  
Size	
  of	
  the	
  token	
  encode	
  size	
  of	
  edit	
  

Tokens	
  are	
  squares	
  	
  
that	
  represent	
  quesFons	
  
SaturaFon	
  encodes	
  Fme	
  	
  

26
Token trajectory and boundaries

Drop	
  Chart	
  

Typo	
  Chart	
  
27
Relax layout and template constraints

Bar	
  chart	
  without	
  boundaries	
  	
  

Pie	
  chart	
  without	
  boundaries	
  	
  
28
Playing with forces: Silo Chart

29
Playing with forces: heap chart

30
Playing with forces: corner chart

31
Aggregation process
Graphical primitive

Token to pixel

Tokens trails
32
Aggregation process: token to pixel

33
Aggregation process: trails

Trails	
  are	
  generated	
  from	
  tokens	
  

Trails	
  are	
  generated	
  from	
  aggregated	
  area	
  
34
Creative mistakes

35
Open questions
How to explore this design space
in a more structured way?
How to interact with aggregated
tokens?
What new designs can be
generated with this approach?
36
Conclusion
Metaphor
Toolkit
Design & parameter space
Exploration
37
Our technic allow

1.	
  to	
  represent	
  incoming	
  data	
  
2.	
  to	
  manage	
  unpredictable	
  update	
  rate	
  	
  	
  	
  
3.	
  to	
  provide	
  context	
  +	
  details	
  over	
  Fme	
  
4.	
  to	
  design	
  a	
  visualizaFons	
  appealing	
  for	
  lay	
  audiences	
  
38
Thanks for your attention !
FEEDBACKS:	
  	
  

@cybunk	
  

MORE	
  INFOS:	
  	
  

www.visualsedimenta0on.org	
  
www.aviz.fr/Research/Huron	
  

FUNDING:	
  	
  
HOSTING:	
  	
  
39
Additional slides

40
Toolkit : Performance
MacBookPro,	
  
Corei7	
  2.4Ghz	
  
8GB	
  Ram,	
  	
  
Mac	
  Os	
  X	
  10.7.5,	
  	
  
Chrome	
  Version	
  
27.0.1	
  
•  Default	
  chart	
  of	
  the	
  
library	
  (a	
  300	
  ×	
  300	
  
pixel	
  canvas	
  with	
  a	
  
3-­‐bin	
  bar	
  chart).	
  	
  
• 
• 
• 
• 
• 

41
Free space,
Boundaries,
Entrance
points,
Forces,
Strata

42

Visual sedimentation - IEEE VIS 2013 Atlanta

  • 1.
    Samuel Huron @cybunk Inria -IRI Romain Vuillemot @romsson Inria (now at Harvard) Jean-Daniel Fekete @jdfaviz Inria 1!
  • 2.
    Data streams areeverywhere! Mail   RSS  feed   Status  update   Logs   2
  • 3.
    Streaming data visualizationchallenges: 1.  How  to  represent  incoming  data?   2.  How  to  manage  unpredictable  update  rate?         3.  How  to  provide  context  +  details  over  Fme?   4.  How  to  design  visualizaFons  for  lay  audiences?   3
  • 4.
  • 5.
    Metaphor: 
 Physical  Mapping   Visual   Data  Mapping   
 from
 physical 
 sedimentation
 Token   Newest   tweet   
 to
 visual 
 sedimentation
 ! Deposit   of  tokens   flocculaFon   Aggregated   Area   Previous   tweets   Older     tweets   5
  • 6.
    We have used the metaphor inreal world situations 6
  • 7.
    We have used the metaphor inreal world situations 7
  • 8.
    We have used the metaphor inreal world situations Average  visit  duraFon  4:21   minutes     Shared  1800  Fmes  on   twi4er       Doubled  the  number  of   tweets.     Good  feedbacks  from  users   8
  • 9.
    We apply thisto classical charts: 9
  • 10.
    And we have generateda lot of “crazy” charts 10
  • 11.
    How it’s done 1.  Design              parameters   2.  Toolkit  and          parameter  space   3.  ExploraFon  of  the            design  space   11
  • 12.
  • 13.
    Free space |Boundaries | Entry points | Forces, Strata 13
  • 14.
    Free space |Boundaries | Entry points | Forces | Strata 14
  • 15.
    Free space |Boundaries | Entry points | Forces | Strata 15
  • 16.
    Free space, Boundaries,Entry points, Forces, Strata Decay   FloccullaFon   Gravity   16
  • 17.
    Free space, Boundaries,Entry points, Forces, Strata 17
  • 18.
    Layout: Templates forclassic charts 18
  • 19.
  • 20.
    How to explore ametaphor-based design space?! In  this  paper,  we..     Create  a  toolkit    and  generate  cases  studies   Defined  a  parameter  space   Generate  a  variety  of  different  designs   Select  “Best”  samples   And  iterate     20
  • 21.
    Toolkit Open  source     on  Github,   GPL  like  license   JavaScript     +  html  5     web  compliant   Box2dWeb  is   used  for  the   physics  engine   D3.js  is  used  for   the  aggregated   area   21
  • 22.
    Creating a chart Apply  it  to  a  DOM  element   var  mySedng  =  {            width:      300,  height:  300,   Canvas  :            chart        {  type:'StackedAreaChart',  …  },   chart:            data          {  model:  [  {label:  'Column  A'},  …  ],   data:            strata      [  [  {initValue:  100,  label:  'Bar  A'}  ],  …  ],   strata:            stream   {   stream:              provider:  'generator',  refresh:    10000/4}            },            sedimenta-on   {   sedimentaFon:                  token:  {  size:  {  original:  6,  minimum:  2  }},                aggregaFon:  {  type:'stacked’  },                suspension:    {  decay:  {power:  1.01  }},                flocculaFon:  {...}            },      };       var  myChart  =    $("#myDivChartContainer”).vs(mySedng);   22
  • 23.
  • 24.
    Let’s play withthis parameter space!! and explore the design space! 24
  • 25.
  • 26.
    Token visual encoding Tokens  are  edits,   Size  of  the  token  encode  size  of  edit   Tokens  are  squares     that  represent  quesFons   SaturaFon  encodes  Fme     26
  • 27.
    Token trajectory andboundaries Drop  Chart   Typo  Chart   27
  • 28.
    Relax layout andtemplate constraints Bar  chart  without  boundaries     Pie  chart  without  boundaries     28
  • 29.
    Playing with forces:Silo Chart 29
  • 30.
    Playing with forces:heap chart 30
  • 31.
    Playing with forces:corner chart 31
  • 32.
  • 33.
  • 34.
    Aggregation process: trails Trails  are  generated  from  tokens   Trails  are  generated  from  aggregated  area   34
  • 35.
  • 36.
    Open questions How toexplore this design space in a more structured way? How to interact with aggregated tokens? What new designs can be generated with this approach? 36
  • 37.
  • 38.
    Our technic allow 1.  to  represent  incoming  data   2.  to  manage  unpredictable  update  rate         3.  to  provide  context  +  details  over  Fme   4.  to  design  a  visualizaFons  appealing  for  lay  audiences   38
  • 39.
    Thanks for yourattention ! FEEDBACKS:     @cybunk   MORE  INFOS:     www.visualsedimenta0on.org   www.aviz.fr/Research/Huron   FUNDING:     HOSTING:     39
  • 40.
  • 41.
    Toolkit : Performance MacBookPro,   Corei7  2.4Ghz   8GB  Ram,     Mac  Os  X  10.7.5,     Chrome  Version   27.0.1   •  Default  chart  of  the   library  (a  300  ×  300   pixel  canvas  with  a   3-­‐bin  bar  chart).     •  •  •  •  •  41
  • 42.