Before we start maps Networks Références
Maps& Networks :
Why and How to Visualize ?
Christophe Bontemps
Toulouse School of Economics, INRA
@Xtophe_Bontemps
Before we start maps Networks Références
WHAT IS DATA VISUALIZATION ?
Data visualisation serves at least two main purposes :
Data exploration
Graphs as visual tests, comparisons (short time to built
and to read)
Before we start maps Networks Références
WHAT IS DATA VISUALIZATION ?
Data visualisation serves at least two main purposes :
Data exploration
Graphs as visual tests, comparisons (short time to built
and to read)
Data representation
Summaries, storytelling (long time to build, short time to
read)
Before we start maps Networks Références
WHAT IS DATA VISUALIZATION ?
Data visualisation serves at least two main purposes :
Data exploration
Graphs as visual tests, comparisons (short time to built
and to read)
Data representation
Summaries, storytelling (long time to build, short time to
read)
The problem is that :
“ Communicating implies simplification
data exploration implies exhaustivity”
Before we start maps Networks Références
SUMMARY OF PREVIOUS EPISODES
From the perspective of the reader, “data visualisation” are
implicitly or explicitly comparisons or even tests (in the
statistical sense).
Graphics should help questioning
Before we start maps Networks Références
SUMMARY OF PREVIOUS EPISODES
From the perspective of the reader, “data visualisation” are
implicitly or explicitly comparisons or even tests (in the
statistical sense).
Graphics should help questioning
They should provide elements, to answer legitimate
questions (or, at least show the data)
Before we start maps Networks Références
SUMMARY OF PREVIOUS EPISODES
From the perspective of the reader, “data visualisation” are
implicitly or explicitly comparisons or even tests (in the
statistical sense).
Graphics should help questioning
They should provide elements, to answer legitimate
questions (or, at least show the data)
If the question implies comparison, they should truthfully
show the comparison
Before we start maps Networks Références
TO CITE A FEW
“All graphs are comparisons” ...
Before we start maps Networks Références
TO CITE A FEW
“All graphs are comparisons” ...
“All of statistics are comparisons (too).” Gelman (2004)
Before we start maps Networks Références
TO CITE A FEW
“All graphs are comparisons” ...
“All of statistics are comparisons (too).” Gelman (2004)
“ The human eye acts is a broad feature detector and general
statistical test”. Buja et al. (2009)
Before we start maps Networks Références
TO CITE A FEW
“All graphs are comparisons” ...
“All of statistics are comparisons (too).” Gelman (2004)
“ The human eye acts is a broad feature detector and general
statistical test”. Buja et al. (2009)
“There are no “good” nor “bad” graphics (...), there are graphics
answering legitimate questions and graphics that do not
answer question at all ” Bertin (1981)
Before we start maps Networks Références
TO CITE A FEW
“All graphs are comparisons” ...
“All of statistics are comparisons (too).” Gelman (2004)
“ The human eye acts is a broad feature detector and general
statistical test”. Buja et al. (2009)
“There are no “good” nor “bad” graphics (...), there are graphics
answering legitimate questions and graphics that do not
answer question at all ” Bertin (1981)
“Above all, show the data” Tufte (2001)
Before we start maps Networks Références
WHAT’S A MAP ?
Visual representation of geographical information and/or
spatial data
A (statistical) map is a summary of data (to be compared
with pictures)
Before we start maps Networks Références
WHAT’S A MAP ?
Visual representation of geographical information and/or
spatial data
A (statistical) map is a summary of data (to be compared
with pictures)
Many types of maps
Before we start maps Networks Références
WHAT’S A MAP ?
Visual representation of geographical information and/or
spatial data
A (statistical) map is a summary of data (to be compared
with pictures)
Many types of maps
What question can ask to a Map ?
Before we start maps Networks Références
WHAT’S A MAP ?
Visual representation of geographical information and/or
spatial data
A (statistical) map is a summary of data (to be compared
with pictures)
Many types of maps
What question can ask to a Map ?
Many rules for drawing a map !
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
Most of the election’s reports are maps : legitimate questions
What happened ?
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
Most of the election’s reports are maps : legitimate questions
What happened ?
How can I see the most important changes ?
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
Most of the election’s reports are maps : legitimate questions
What happened ?
How can I see the most important changes ?
Which are the states that vote the most for Clinton ? For
Trump ?
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
Most of the election’s reports are maps : legitimate questions
What happened ?
How can I see the most important changes ?
Which are the states that vote the most for Clinton ? For
Trump ?
Can I rank the importance of each state in the outcome ?
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
2016 results as a map : Washington Post
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
2012 results as a map : New York Time
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
2008 results as a map : New York Time
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
What’s wrong ? I cannot see anything ! I cannot answer
legitimate questions !
Quantitative information is over represented in "large"
states
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
What’s wrong ? I cannot see anything ! I cannot answer
legitimate questions !
Quantitative information is over represented in "large"
states
Easy to see the difference between states
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
What’s wrong ? I cannot see anything ! I cannot answer
legitimate questions !
Quantitative information is over represented in "large"
states
Easy to see the difference between states
Easy to see the results for one state
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
What’s wrong ? I cannot see anything ! I cannot answer
legitimate questions !
Quantitative information is over represented in "large"
states
Easy to see the difference between states
Easy to see the results for one state
Easy to see "regions" of the same colour
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
What’s wrong ? I cannot see anything ! I cannot answer
legitimate questions !
Quantitative information is over represented in "large"
states
Easy to see the difference between states
Easy to see the results for one state
Easy to see "regions" of the same colour
Only the "raw" elements are visible, not the details (limited
questions)
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
Maybe a better map ? From Financial Times blog
From Financial Times
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
2012 results as a map : New York Time
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
Maybe spatial information is not the most relevant !
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
Maybe spatial information is not the most relevant !
Before we start maps Networks Références
A CASE STUDY : ELECTIONS
Or if really geography matters (2004 campaign) :
From Andrew Gelman’s Blog. Each pixel represents 1,000 votes.
Before we start maps Networks Références
GEOGRAPHY EXISTS IN THE DATA, BUT IS NOT THE
MATTER OF COMPARISON
Before we start maps Networks Références
ANOTHER GOOD EXAMPLE FOR NOT USING A MAP
From Our World in Data
Before we start maps Networks Références
EXAMPLES OF MAPS MISSISSIPPI MEANDERS
The interest here is to
compare location, lines,
Before we start maps Networks Références
AIR MAP OF THE WORLD (1943)
Before we start maps Networks Références
THE LONDON COUNTY COUNCIL BOMB DAMAGE
MAPS, 1939-1945
Before we start maps Networks Références
AIR ROUTE MAPS : AARON KOBLIN
Before we start maps Networks Références
PARTIAL CONCLUSION
Maps are useful
Before we start maps Networks Références
PARTIAL CONCLUSION
Maps are useful
There are rules (over representation of large area,
colorpleth,etc..)
Before we start maps Networks Références
PARTIAL CONCLUSION
Maps are useful
There are rules (over representation of large area,
colorpleth,etc..)
Statistical maps are difficult to built :
Before we start maps Networks Références
PARTIAL CONCLUSION
Maps are useful
There are rules (over representation of large area,
colorpleth,etc..)
Statistical maps are difficult to built :
How to show uncertainty on a map ?
Before we start maps Networks Références
PARTIAL CONCLUSION
Maps are useful
There are rules (over representation of large area,
colorpleth,etc..)
Statistical maps are difficult to built :
How to show uncertainty on a map ?
How to show correlations on a map ?
Before we start maps Networks Références
PARTIAL CONCLUSION
Maps are useful
There are rules (over representation of large area,
colorpleth,etc..)
Statistical maps are difficult to built :
How to show uncertainty on a map ?
How to show correlations on a map ?
Comparisons are limited (only geographical)
Before we start maps Networks Références
PARTIAL CONCLUSION
Maps are useful
There are rules (over representation of large area,
colorpleth,etc..)
Statistical maps are difficult to built :
How to show uncertainty on a map ?
How to show correlations on a map ?
Comparisons are limited (only geographical)
Maps can help select, or highlight other elements (see
D3.js, dashboards, or GeoXP package)
Before we start maps Networks Références
QUESTIONS ON NETWORKS :
There are many challenges in statistical networks analysis :
Statistics on networks are quite difficult to handle
Before we start maps Networks Références
QUESTIONS ON NETWORKS :
There are many challenges in statistical networks analysis :
Statistics on networks are quite difficult to handle
Many types of networks
Before we start maps Networks Références
QUESTIONS ON NETWORKS :
There are many challenges in statistical networks analysis :
Statistics on networks are quite difficult to handle
Many types of networks
Questions related to networks are hard to state (fuzzy or
unarticulated)
Before we start maps Networks Références
QUESTIONS ON NETWORKS :
There are many challenges in statistical networks analysis :
Statistics on networks are quite difficult to handle
Many types of networks
Questions related to networks are hard to state (fuzzy or
unarticulated)
Visualizations are barely readable (decoding is hard !)
Before we start maps Networks Références
IS THIS A NETWORK OR A MAP ?
Before we start maps Networks Références
IS THIS A NETWORK OR A MAP ?
This is a Map showing links (air connections) between
geographical entities.
Before we start maps Networks Références
IS THIS A NETWORK OR A MAP ?
This is a Map showing links (air connections) between
geographical entities.
This is “a collection of interconnected things”, (Oxford Englis
Dictionary)
Before we start maps Networks Références
TYPES OF NETWORKS
Adjacency Networks
Before we start maps Networks Références
TYPES OF NETWORKS
Adjacency Networks
Before we start maps Networks Références
TYPES OF NETWORKS
Adjacency Networks
Bipartite Networks
Before we start maps Networks Références
TYPES OF NETWORKS
Adjacency Networks
Bipartite Networks
Before we start maps Networks Références
TYPES OF NETWORKS
Adjacency Networks
Bipartite Networks
From bipartite, one can create adjacency
Before we start maps Networks Références
TYPES OF NETWORKS
Adjacency Networks
Bipartite Networks
From bipartite, one can create adjacency
Before we start maps Networks Références
TYPES OF NETWORKS
Adjacency Networks
Bipartite Networks
From bipartite, one can create adjacency
From Heijmans et al. (2014), and Kolaczyk (2009)
Before we start maps Networks Références
TYPES OF NETWORKS
Undirected (A& B ’know each other’) and directed (A ’owe
money to’ B)
Before we start maps Networks Références
TYPES OF NETWORKS
Undirected (A& B ’know each other’) and directed (A ’owe
money to’ B)
Before we start maps Networks Références
TYPES OF NETWORKS
Undirected (A& B ’know each other’) and directed (A ’owe
money to’ B)
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
FAMILIES OF GRAPHS
Complete (top left) ; ring (top right) ; tree (bottom left) ; and
star (bottom right)
Before we start maps Networks Références
FAMILIES OF GRAPHS
Complete (top left) ; ring (top right) ; tree (bottom left) ; and
star (bottom right)
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
Relationships of all of Victor Hugo’s characters of "Les
Miserables".
Data composed of nodes (characters) + Links (strength of
links)
http://bl.ocks.org/mbostock/4062045
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
Relationships of all of Victor Hugo’s characters of "Les
Miserables".
Data composed of nodes (characters) + Links (strength of
links)
Nodes can be coloured (information on characters)
http://bl.ocks.org/mbostock/4062045
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
Relationships of all of Victor Hugo’s characters of "Les
Miserables".
Data composed of nodes (characters) + Links (strength of
links)
Nodes can be coloured (information on characters)
Force-directed graph
http://bl.ocks.org/mbostock/4062045
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
Relationships of all of Victor Hugo’s characters of "Les
Miserables".
Data composed of nodes (characters) + Links (strength of
links)
Nodes can be coloured (information on characters)
Force-directed graph
Here, done with d3.js
http://bl.ocks.org/mbostock/4062045
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
Relationships of all of Victor Hugo’s characters of "Les
Miserables".
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
What do we see ? An overview of problems :
Not the same graph each time (nondeterminstic algorithm)
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
What do we see ? An overview of problems :
Not the same graph each time (nondeterminstic algorithm)
Spacial position does not encode attributes
(algorithm-driven)
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
What do we see ? An overview of problems :
Not the same graph each time (nondeterminstic algorithm)
Spacial position does not encode attributes
(algorithm-driven)
Proximity (of nodes) sometimes meaningful sometimes
arbitrary. Artefacts created (some nodes “pushed back”.)
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
What do we see ? An overview of problems :
Not the same graph each time (nondeterminstic algorithm)
Spacial position does not encode attributes
(algorithm-driven)
Proximity (of nodes) sometimes meaningful sometimes
arbitrary. Artefacts created (some nodes “pushed back”.)
What can we learn ? What can we compare ?
Before we start maps Networks Références
A CLASSICAL NETWORK EXAMPLE
What do we see ? An overview of problems :
Not the same graph each time (nondeterminstic algorithm)
Spacial position does not encode attributes
(algorithm-driven)
Proximity (of nodes) sometimes meaningful sometimes
arbitrary. Artefacts created (some nodes “pushed back”.)
What can we learn ? What can we compare ?
Alternative view with bezier
http://bl.ocks.org/mbostock/4600693
Before we start maps Networks Références
HOW DOES IT WORK ?
A vast family of Force-directed placement network
representation
Placing nodes randomly (put colour on groups)
Before we start maps Networks Références
HOW DOES IT WORK ?
A vast family of Force-directed placement network
representation
Placing nodes randomly (put colour on groups)
Iteratively refine location following “forces”
Before we start maps Networks Références
HOW DOES IT WORK ?
A vast family of Force-directed placement network
representation
Placing nodes randomly (put colour on groups)
Iteratively refine location following “forces”
Nodes have “repulsive forces” (like electrons)
Before we start maps Networks Références
HOW DOES IT WORK ?
A vast family of Force-directed placement network
representation
Placing nodes randomly (put colour on groups)
Iteratively refine location following “forces”
Nodes have “repulsive forces” (like electrons)
Links are “springs” (strength proportional to nb of
connections)
Before we start maps Networks Références
HOW DOES IT WORK ?
A vast family of Force-directed placement network
representation
Placing nodes randomly (put colour on groups)
Iteratively refine location following “forces”
Nodes have “repulsive forces” (like electrons)
Links are “springs” (strength proportional to nb of
connections)
Apply “some other rules‘” (see next slides)
Before we start maps Networks Références
HOW DOES IT WORK ?
A vast family of Force-directed placement network
representation
Placing nodes randomly (put colour on groups)
Iteratively refine location following “forces”
Nodes have “repulsive forces” (like electrons)
Links are “springs” (strength proportional to nb of
connections)
Apply “some other rules‘” (see next slides)
Iterate (computationally demanding)
Before we start maps Networks Références
HOW DOES IT WORK ?
A vast family of Force-directed placement network
representation
Placing nodes randomly (put colour on groups)
Iteratively refine location following “forces”
Nodes have “repulsive forces” (like electrons)
Links are “springs” (strength proportional to nb of
connections)
Apply “some other rules‘” (see next slides)
Iterate (computationally demanding)
Wait until it stabilizes (even on a local equilibrium)
Before we start maps Networks Références
FORCE-DIRECTED GRAPHS :
Physically-based model + aesthetic touch + fuzzy stuff...
From Tunkelang (1998)
Before we start maps Networks Références
RULES FOR “NODE-LINK” REPRESENTATIONS
Node-links representation follow rules :
General rules :
From Katya Ognyanova (2015) and Bahoken et al. (2013)
Before we start maps Networks Références
RULES FOR “NODE-LINK” REPRESENTATIONS
Node-links representation follow rules :
General rules :
Most connected nodes in the center
From Katya Ognyanova (2015) and Bahoken et al. (2013)
Before we start maps Networks Références
RULES FOR “NODE-LINK” REPRESENTATIONS
Node-links representation follow rules :
General rules :
Most connected nodes in the center
Less connected nodes at the periphery
From Katya Ognyanova (2015) and Bahoken et al. (2013)
Before we start maps Networks Références
RULES FOR “NODE-LINK” REPRESENTATIONS
Node-links representation follow rules :
General rules :
Most connected nodes in the center
Less connected nodes at the periphery
Layout aesthetics rules :
From Katya Ognyanova (2015) and Bahoken et al. (2013)
Before we start maps Networks Références
RULES FOR “NODE-LINK” REPRESENTATIONS
Node-links representation follow rules :
General rules :
Most connected nodes in the center
Less connected nodes at the periphery
Layout aesthetics rules :
Minimize edge crossing
From Katya Ognyanova (2015) and Bahoken et al. (2013)
Before we start maps Networks Références
RULES FOR “NODE-LINK” REPRESENTATIONS
Node-links representation follow rules :
General rules :
Most connected nodes in the center
Less connected nodes at the periphery
Layout aesthetics rules :
Minimize edge crossing
Prevent overlap
From Katya Ognyanova (2015) and Bahoken et al. (2013)
Before we start maps Networks Références
RULES FOR “NODE-LINK” REPRESENTATIONS
Node-links representation follow rules :
General rules :
Most connected nodes in the center
Less connected nodes at the periphery
Layout aesthetics rules :
Minimize edge crossing
Prevent overlap
Symmetry
From Katya Ognyanova (2015) and Bahoken et al. (2013)
Before we start maps Networks Références
RULES FOR “NODE-LINK” REPRESENTATIONS
Node-links representation follow rules :
General rules :
Most connected nodes in the center
Less connected nodes at the periphery
Layout aesthetics rules :
Minimize edge crossing
Prevent overlap
Symmetry
Uniform edge length
From Katya Ognyanova (2015) and Bahoken et al. (2013)
Before we start maps Networks Références
LAYOUT AESTHETICS
From Katya Ognyanova (2015)
Before we start maps Networks Références
A NOTE ON DATA : MATRIX
From Katya Ognyanova (2015)
Before we start maps Networks Références
A NOTE ON DATA : EDGE LIST
From Katya Ognyanova (2015)
Before we start maps Networks Références
A NOTE ON DATA : NODE LIST
From Katya Ognyanova (2015)
Before we start maps Networks Références
A NOTE ON DATA : NODE ATTRIBUTES
From Katya Ognyanova (2015)
Before we start maps Networks Références
ALTERNATIVE VIEW : ADJACENT MATRIX PLOT
An adjacency matrix, where each cell ij represents an edge from
vertex i to vertex j. Here, vertices represent characters in a
book, while edges represent co-occurrence in a chapter.
http://bost.ocks.org/mike/miserables/
Before we start maps Networks Références
NETWORKS : ADJACENT MATRIX PLOT
As in many example, sorting is very useful !
Before we start maps Networks Références
NETWORKS : ADJACENT MATRIX PLOT
Adjacent matrix help identifying patterns
From McGuffin (2012)
Before we start maps Networks Références
NETWORKS : ARC DIAGRAM PLOT
To put some structure in the graph : Horizontal alignment
From Gaston Sanchez see also McGuffin (2012)
Before we start maps Networks Références
NETWORKS : ARC DIAGRAM PLOT
Now we can sort ! (according to groups, and then degrees)
FromGaston Sanchez
Before we start maps Networks Références
NETWORKS : ARC DIAGRAM PLOT : MY TOUCH !
Before we start maps Networks Références
NETWORKS : ROLE OF TOPOLOGY
Circular layouts of a 43-node, 80-edge network. Force-directed
graph
From McGuffin (2012)
Before we start maps Networks Références
NETWORKS : ADDING A TOPOLOGY HELPS
Circular layouts of a 43-node, 80-edge network. No ordering,
with curved edges.
From McGuffin (2012)
Before we start maps Networks Références
NETWORKS : ADDING A TOPOLOGY HELPS
Circular layouts of a 43-node, 80-edge network. Barycenter
ordering, with curved edges.
Before we start maps Networks Références
PROBLEM 1 : IDENTIFY REFERENCE GRAPHS
Random Graphs difficult to identify
Before we start maps Networks Références
PROBLEM 1 : IDENTIFY REFERENCE GRAPHS
Random Graphs difficult to identify
Before we start maps Networks Références
PROBLEM 1 : IDENTIFY REFERENCE GRAPHS
Random Graphs difficult to identify
Or
Before we start maps Networks Références
PROBLEM 1 : IDENTIFY REFERENCE GRAPHS
Random Graphs difficult to identify
Or
Before we start maps Networks Références
PROBLEM 1 : IDENTIFY REFERENCE GRAPHS
Random Graphs difficult to identify
Or
→ How to do any visual test ?
Before we start maps Networks Références
PROBLEM 2 : LAYOUT CHANGE PERCEPTION
Three layouts of the same graph
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
PROBLEM 2 : LAYOUT CHANGE PERCEPTION
Three layouts of the same graph
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
PROBLEM 2 : LAYOUT CHANGE PERCEPTION
Three layouts of the same graph
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
PROBLEM 2 : LAYOUT CHANGE PERCEPTION
Three layouts of the same graph
→ Circular, radial, layered.
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
PROBLEM 3 : DECORATION HELPS ?
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
PROBLEM 3 : DECORATION HELPS ?
Plain and decorated visualizations of a karate club network
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
PROBLEM 4 : SCALABILITY ?
(192 blogs linked by 1,431 edges indicating at least one of the two blogs referenced the other). From Kolaczyk and
Csárdi (2014)
Before we start maps Networks Références
PROBLEM 4 : SCALABILITY ?
French political blog network
(192 blogs linked by 1,431 edges indicating at least one of the two blogs referenced the other). From Kolaczyk and
Csárdi (2014)
Before we start maps Networks Références
PROBLEM 4 : SCALABILITY ?
French political blog network
Towards an "hairball" + complexity is O(N2
v)
(192 blogs linked by 1,431 edges indicating at least one of the two blogs referenced the other). From Kolaczyk and
Csárdi (2014)
Before we start maps Networks Références
SOLUTIONS : SPLIT ?
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
SOLUTIONS : SPLIT ?
Two ego-centric views of the karate club network
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
SOLUTIONS : AGGREGATE ?
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
SOLUTIONS : AGGREGATE ?
French political blog network at the level of political
parties
From Kolaczyk and Csárdi (2014)
Before we start maps Networks Références
REFERENCES I
Bahoken, F., Beauguitte, L., and Lhomme, S. (2013). La visualisation des
réseaux. principes, enjeux et perspectives.
Bertin, J. (1981). Théorie matricielle de la graphique. Communication et
langages, 48(1) :62–74.
Buja, A., Cook, D., Hofmann, H., Lawrence, M., Lee, E.-K., Swayne, D. F., and
Wickham, H. (2009). Statistical inference for exploratory data analysis and
model diagnostics. Philosophical Transactions of the Royal Society of London
A : Mathematical, Physical and Engineering Sciences, 367(1906) :4361–4383.
Gelman, A. (2004). Exploratory data analysis for complex models. Journal of
Computational and Graphical Statistics, 13(4).
Heijmans, R., Heuver, R., Levallois, C., and Lelyveld, I. V. (2014). Dynamic
visualization of large transaction networks : the daily dutch overnight
money market.
Kolaczyk, E. D. (2009). Statistical Analysis of Network Data : Methods and
Models. Springer Series in Statistics. Springer-Verlag New York, 1 edition.
Kolaczyk, E. D. and Csárdi, G. (2014). Statistical Analysis of Network Data with
R. Use R ! 65. Springer-Verlag New York, 1 edition.
Before we start maps Networks Références
REFERENCES II
McGuffin, M. J. (2012). Simple algorithms for network visualization : A
tutorial. Tsinghua Science and Technology, 17(4) :383–398.
Ognyanova, K. (2015). R network visualization workshop. In POLNET 2015,
Portland OR.
Tufte, E. R. (2001). The Visual Display of Quantitative Information. Graphics
Press, 2 edition.
Tunkelang, D. (1998). A Numerical Optimization Approach to Graph Drawing.
PhD thesis, Dissertation, Carnegie Mellon University, School of Computer
Science.