Skip to content

Instantly share code, notes, and snippets.

@nbremer
Last active June 12, 2016 08:22
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save nbremer/7658623 to your computer and use it in GitHub Desktop.
Save nbremer/7658623 to your computer and use it in GitHub Desktop.
Chord diagram - Switching behavior between phone brands of the Dutch
height: 1500

The full blog about the 2013 version of the Chord Diagram created for the Global Mobile Consumer Survey held by Deloitte each year. It was featured on the Deloitte Netherlands webpage. It shows the flows of people between the phone brand of their previous phone and the current phone

For an explanation on how to understand this chart, please take a look at my "Storytelling" version of the same Chord Diagram

My blog about this Chord Diagram can be found here

<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html;charset=utf-8"/>
<title>Chord Diagram - Phone Brand Switching Behavior</title>
<script src="http://d3js.org/d3.v3.min.js"></script>
<!-- <link type="text/css" rel="stylesheet" href="style.css"/> -->
<style>
body {
/*overflow: hidden;*/
margin: 0;
font-size: 14px;
font-family: "Helvetica Neue", Helvetica;
}
#footer {
position: absolute;
top: 0;
z-index: 1;
display: block;
font-size: 36px;
font-weight: 300;
text-shadow: 0 1px 0 #fff;
top: 30px;
left: 30px;
text-align: left;
}
#owner{
position: absolute;
z-index: 1;
display: block;
font-weight: 300;
text-shadow: 0 1px 0 #fff;
top: 1500px;
left: 30px;
text-align: left;
color: #474747;
padding-bottom: 20px;
font-size: 10px;
}
#SideText{
position: absolute;
display: block;
font-size: 12px;
color: #474747;
top: 170px;
left: 950px;
width: 200px;
}
#BelowText{
position: absolute;
font-size: 14px;
color: #474747;
top: 900px;
left: 30px;
width: 1000px;
}
line {
stroke: #000;
stroke-width: 1.px;
}
.hint {
position: absolute;
left: 0;
width: 500px;
font-size: 12px;
color: #999;
}
text {
font-size: 10px;
}
.titles{
font-size: 14px;
}
.annotate{
font-size: 11px;
fill: #474747;
}
path.chord {
fill-opacity: .80;
}
.pTitle {
font-size: 16px;
text-shadow: 0 1px 0 #fff;
fill: #474747;
font-weight: bold;
}
.survey {
font-size: 10px;
}
</style>
</head>
<body>
<div id="body">
<div id="footer">
Switching behavior between phone brands of the Dutch
<div class="hint">
<span>Analyzing the results of the Deloitte Mobile Consumer Survey<span/></br>
<span>Mouse over a brand to highlight</span>
</div>
</div>
<div id="SideText">
<p id="p0" class="pTitle Side">Chord Diagram</p>
<p id="p1" class="Side">This chord diagram shows relationships in terms of switching behaviour among phone brands.</p>
<p id="p2" class="Side">The chords are directed: for example, while 12% of respondents who now have a Samsung used to have a Nokia, only 2% have the opposite; 2% now have a Nokia and used to have a Samsung phone.</p>
<p id="p3" class="Side">The thickness of chords between brands encodes the percentage of respondents that switched between two brands: thicker chords represent a higher percentage of respondents that made that switch.</p>
<p id="p4" class="Side">This diagram reveals asymmetries: if a chord is tapered, there are more respondents switching to one brand from the other than vice versa.</p>
<p id="p5" class="Side">Mouse over a brand name to focus on switching behaviour to and from other brands concerning that specific brand.</p>
</div>
<div id="chart"></div>
<div id="BelowText">
<p>
While almost everyone in the Netherlands has a phone, many consumers switch to a new phone whenever their contract allows it. But what are the reasons for this switch?
</p>
<p>
The chord diagram is a relatively new and powerful visualization, especially when wanting to show and analyse connections. However, it does require some explanation for those who have never seen one before. We have analysed the connection between the respondent's current and previous phone brand through this chord diagram to explain precisely that.
</p>
<p>
The chord diagram represents the installed base shares (all phones) of different brands and the switching behaviour of customers between these brands.
</p>
<p>
The circle is split up into 8 brands with the arc length of each group scaled to the current brand’s market share. The percentages along the outer rim of the chord Diagram give the share-per-brand. This shows that Samsung dominates the installed base with a 41% share, Nokia is a surprising second with 20%, and Apple is third with 13%.
</p>
<p>
The chords between the arcs visualize the switching behaviour of the respondents between brands in both directions. For example, the big dark blue chord connecting Samsung and Nokia in the lower left section shows the part of respondents that moved from Samsung to Nokia and from Nokia to Samsung. Clearly, the figure shows that Nokia has lost its installed base share to Samsung, as 12% of all respondents that used to own a Nokia now own a Samsung. At the same time, only 2 % of the Nokia owners previously owned a Samsung.
</p>
<p>
To clarify which brand has been the net gainer, each chord has the colour of the brand that managed to gain more customers from the connected brand than it has lost to this brand. With this in mind, we can immediately see that Apple is a true net gainer; all its chords are 'Apple grey'. Even more so, besides the chord to Samsung, all chords at Apple are extremely thin at the other side, thus Apple has lost virtually no customers to other brands except Samsung.
</p>
<p>
There are also chords that connect each brand to itself. These ‘hill-like’ chords represent the respondents who did not switch brands, but stayed loyal to their brand.
</p>
<p>
These examples, used to explain the chord diagram, are but a handful of the interesting facts that can be gained from it. I encourage you to study the chord diagram by yourself to find all insights in can provide and play with the online interactive version.
</p>
<p>
Finally, this analysis is but a small piece of the total report based on the yearly Global Mobile Consumer Survey that Deloitte performed. Please feel free to contact the team members for more information regarding the complete report and also how you can discuss this with your clients
</p>
<p class="survey"><b>Survey</b></br>
Deloitte has created the Global Mobile Consumer Survey – an annual snapshot of the mobile consumer, based on insights from more than 30.000 respondents in more than 20 countries.
This analysis is based on the Dutch mobile market anno 2013, based on over 2000 respondents from the Netherlands, where this sample is nationally representative.
All research has been undertaken via online research. Fieldwork took place between May to June 2013.
The questions for this survey were written by Deloitte member firms. The multinational online research program was managed by Ipsos MORI.</p>
<p class="survey"><b>Project team</b></br>
Stephen Ward (sponsoring partner), Gagandeep Sethi (Project Manager), Frank Korf (Analytics lead), Arseni Storojev, Nadieh Bremer, Jessica Abad Kelly
</p>
</div>
<div id="owner">Analysis and Visualization created by <A HREF="http://www.visualcinnamon.com/">Nadieh Bremer</A></div>
</div>
<script type="text/javascript">
var NameProvider = ["Apple","Blackberry","HTC","LG","Nokia","Samsung","Sony","Other"];
var matrix = [
[5.0858,1.777,0.7353,0.4902,2.6348,1.6544,0.5515,0.4289], //Apple
[0.0613,1.2868,0.2451,0,0.7353,0.6127,0.1225,0], //Blackberry
[0.1225,0.3676,2.5735,0.1225,2.2672,1.8382,0.7353,0.0613], //HTC
[0.0613,0.1225,0.2451,0.6127,1.348,0.9804,0.3676,0.5515], //LG
[0.2451,0.3064,0.1838,0.6127,14.951,1.8382,0.4289,1.777], //Nokia
[1.4706,1.8382,1.777,2.3897,12.2549,17.8922,2.2059,1.7157], //Samsung
[0.0613,0.2451,0.5515,0.2451,1.348,0.7966,1.777,0.3064], //Sony
[0,0,0.1225,0.3064,1.1642,0.6127,0.4902,1.2868] //Other
];
var colors = ["#C4C4C4","#000000","#69B40F","#C8125C","#008FC8","#10218B","#134B24","#737373"];
var chord = d3.layout.chord()
.padding(.04)
.sortSubgroups(d3.descending) //sort the chords inside an arc from high to low
.sortChords(d3.descending) //which chord should be shown on top when chords cross. Now the biggest chord is at the bottom
.matrix(matrix);
var width = 900,
height = 650,
innerRadius = Math.min(width, height) * .39,
outerRadius = innerRadius * 1.09;
var fill = d3.scale.ordinal()
.domain(d3.range(NameProvider.length))
.range(colors);
var arc = d3.svg.arc()
.innerRadius(innerRadius)
.outerRadius(outerRadius);
var svg = d3.select("#chart").append("svg:svg")
.attr("width", width)
.attr("height", height+250)
.append("svg:g")
.attr("transform", "translate(" + (width / 2 + 30) + "," + (height / 2+180) + ")");
////////////////////////////////////////////////////////////
////////////////// Draw outer Arcs /////////////////////////
////////////////////////////////////////////////////////////
var g = svg.selectAll("g.group")
.data(chord.groups)
.enter().append("svg:g")
.attr("class", "group")
.on("mouseover", fadeIn(.02))
.on("mouseout", fadeOut(.80));
g.append("svg:path")
.style("stroke", function(d) { return fill(d.index); })
.style("fill", function(d) { return fill(d.index); })
.attr("d", arc);
////////////////////////////////////////////////////////////
////////////////// Append Ticks ////////////////////////////
////////////////////////////////////////////////////////////
var ticks = svg.append("svg:g").selectAll("g.ticks")
.data(chord.groups)
.enter().append("svg:g").selectAll("g.ticks")
.attr("class", "ticks")
.data(groupTicks)
.enter().append("svg:g")
.attr("transform", function(d) {
return "rotate(" + (d.angle * 180 / Math.PI - 90) + ")"
+ "translate(" + outerRadius+40 + ",0)";
});
ticks.append("svg:line")
.attr("x1", 1)
.attr("y1", 0)
.attr("x2", 5)
.attr("y2", 0)
.style("stroke", "#000");
ticks.append("svg:text")
.attr("x", 8)
.attr("dy", ".35em")
.attr("transform", function(d) { return d.angle > Math.PI ? "rotate(180)translate(-16)" : null; })
.style("text-anchor", function(d) { return d.angle > Math.PI ? "end" : null; })
.text(function(d) { return d.label; });
////////////////////////////////////////////////////////////
////////////////// Append Names ////////////////////////////
////////////////////////////////////////////////////////////
g.append("svg:text")
.each(function(d) { d.angle = (d.startAngle + d.endAngle) / 2; })
.attr("dy", ".35em")
.attr("class", "titles")
.attr("text-anchor", function(d) { return d.angle > Math.PI ? "end" : null; })
.attr("transform", function(d) {
return "rotate(" + (d.angle * 180 / Math.PI - 90) + ")"
+ "translate(" + (innerRadius + 55) + ")"
+ (d.angle > Math.PI ? "rotate(180)" : "");
})
.text(function(d,i) { return NameProvider[i]; });
////////////////////////////////////////////////////////////
////////////////// Draw inner chords ///////////////////////
////////////////////////////////////////////////////////////
svg.selectAll("path.chord")
.data(chord.chords)
.enter().append("svg:path")
.attr("class", "chord")
.style("stroke", function(d) { return d3.rgb(fill(d.source.index)).darker(); })
.style("fill", function(d) { return fill(d.source.index); })
.attr("d", d3.svg.chord().radius(innerRadius));
////////////////////////////////////////////////////////////
////////////////////// Annotate ////////////////////////////
////////////////////////////////////////////////////////////
//From Nokia to Samsung
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", -445)
.attr("y", 10)
.text("12% of respondents switched");
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", -445)
.attr("y", 23)
.text("from Nokia to Samsung");
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", -445)
.attr("y", 38)
.text("That is more than half of");
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", -445)
.attr("y", 51)
.text("those currently at Nokia");
svg.append("svg:line")
.attr("class", "annotate")
.attr("x1", -445)
.attr("y1", -3)
.attr("x2", -290)
.attr("y2", -3)
.style("stroke", "#858585");
//From Samsung to Nokia
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", 90)
.attr("y", 340)
.text("Only 2% of respondents");
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", 90)
.attr("y", 353)
.text("switched from Samsung to Nokia");
svg.append("svg:line")
.attr("class", "annotate")
.attr("x1", 85)
.attr("y1", 275)
.attr("x2", 85)
.attr("y2", 354)
.style("stroke", "#858585");
//Nokia Loyal
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", 420)
.attr("y", 130)
.attr("text-anchor","end")
.text("15% of respondents");
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", 420)
.attr("y", 143)
.attr("text-anchor","end")
.text("stayed loyal to Nokia");
svg.append("svg:line")
.attr("class", "annotate")
.attr("x1", 265)
.attr("y1", 117)
.attr("x2", 420)
.attr("y2", 117)
.style("stroke", "#858585");
//Apple
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", 63)
.attr("y", -383)
.text("Because all chords at the arc of Apple have");
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", 63)
.attr("y", -370)
.text("the color of Apple (grey), Apple is taking more clients");
svg.append("svg:text")
.attr("class", "annotate")
.attr("x", 63)
.attr("y", -357)
.text("from all other brands than it is losing to them");
svg.append("svg:line")
.attr("class", "annotate")
.attr("x1", 58)
.attr("y1", -282)
.attr("x2", 58)
.attr("y2", -393)
.style("stroke", "#858585");
////////////////////////////////////////////////////////////
////////////////// Extra Functions /////////////////////////
////////////////////////////////////////////////////////////
// Returns an event handler for fading a given chord group.
function fadeIn(opacity) {
return function(d, i) {
d3.selectAll(".Side")
.transition()
.duration(200)
.style("opacity", 0.0001);
d3.selectAll(".annotate")
.transition()
.duration(200)
.style("opacity", 0.0001);
d3.select("#p0").transition().duration(200).delay(200).style("opacity", 1).text(NameProvider[i]);
if(i == 0) { //Apple
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("All chords connected to Apple are ‘Apple grey’, thus showing that Apple has been able to take more customers from all other brands than vice versa, making it the most successful net gainer of all brands in this analysis");
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text("Additionally, more than a third of the current iPhone owners also previously had an iPhone, implying significant loyalty among Apple owners. This is in line with the finding that around one fifth of Apple owners responded to have bought their device because they have other devices of the same brand, like iPad or iPod. ");
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text("The respondents having an iPhone pay significantly more attention than other brand owners to purchasing criteria like the handset brand, design, new model on the market, broad range of available apps, streaming content to other devices and the fact that most of their friends have the same device. ");
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text(" ");
} else if (i == 1) { //Blackberry
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("Blackberry is the smallest brand not yet grouped into the 'Other' category with a share of 3% among respondents.");
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text("Almost half of their current base is loyal to Blackberry; their previous phone was a Blackberry as well.");
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text("However, it appears that Blackberry lost a large share (3%) to Apple and Samsung (the chord ending at Samsung and Apple are much wider than the beginning of these chords at Blackberry.)");
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text("Blackberry also lost more customers to all other brands than it took from them, except for Nokia");
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text(" ");
} else if (i == 2) { //HTC
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("HTC is especially successful in taking customers from Nokia, but it seems to take as many customers from Samsung, Sony and LG as it lost to them (the chords connecting HTC with these three brands begin and end in similar sizes).");
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text(" ");
} else if (i == 3) { //LG
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("Slightly larger than the smallest brand not yet in the 'Other' group, LG is currently owned by 4% of the respondents.");
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text("LG is taking more clients from all other brands, except Apple and Samsung, than it is losing to them. For those chords in the color of LG, only the chord connecting Nokia and LG is quite asymetrical. LG is therefore better in taking clients from Nokia than vice versa.");
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text("The brand that is most successful in taking customers away from LG is Samsung; 2% of respondents are now at Samsung and came from LG - A significant portion of the share currently at LG");
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text(" ");
} else if (i == 4) { //Nokia
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("Nokia is still the second-largest brand in the mobile phone installed base with a 20% share among respondents. ");
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text("A very large fraction of these phones belong to the feature phone category. Nokia is a net loser compared to all other main brands. The most dramatic loss is to Samsung; when comparing sizes, Nokia lost more than half of its current share to Samsung alone. ");
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text("Compared to other brand owners, Nokia owners were particularly interested in the device price when purchasing their latest devices, besides factors applicable to all brands like reliability.");
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text(" ");
} else if (i == 5) { //Samsung
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("Samsung is dominating the mobile phone installed base, with more than 40% of respondents currently using a Samsung phone. ");
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text("Besides being a net gainer from most other brands, Samsung has also been very successful in keeping their customers loyal; almost half of the current Samsung share had a Samsung as their previous phone. ");
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text("According to our survey results, Samsung owners were paying more attention to purchasing criteria like device price, cost of tariff, the quality of touchscreen, phone size and battery life than Apple users. These factors could be the most important differentiators attributing to the success of Samsung in the broad mass market.");
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text(" ");
} else if (i == 6) { //Sony
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("Sony appears to lose as many customers to (most) other brands, as it is taking from them. Only Samsung is able to take a significant bigger portion from Sony than vice versa.");
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text(" ");
} else if (i == 7) { //Other
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("Brands that have been grouped in the 'Other' category include; Motorola, Huawei and ZTE");
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text(" ");
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text(" ");
};
svg.selectAll("path.chord")
.filter(function(d) { return d.source.index != i && d.target.index != i; })
.transition()
.style("stroke-opacity", opacity)
.style("fill-opacity", opacity);
};
}//fadeIn
// Returns an event handler for fading a given chord group.
function fadeOut(opacity) {
return function(d, i) {
d3.selectAll(".Side")
.transition()
.duration(200)
.style("opacity", 0.0001);
d3.selectAll(".annotate")
.transition()
.duration(200)
.style("opacity", 1);
d3.select("#p0").transition().duration(200).delay(200).style("opacity", 1).text("Chord Diagram")
d3.select("#p1").transition().duration(200).delay(200).style("opacity", 1).text("This chord diagram shows relationships in terms of switching behaviour among phone brands.")
d3.select("#p2").transition().duration(200).delay(200).style("opacity", 1).text("The chords are directed: for example, while 12% of respondents who now have a Samsung used to have a Nokia, only 2% have the opposite; 2% now have a Nokia and used to have a Samsung phone.")
d3.select("#p3").transition().duration(200).delay(200).style("opacity", 1).text("The thickness of chords between brands encodes the percentage of respondents that switched between two brands: thicker chords represent a higher percentage of respondents that made that switch.")
d3.select("#p4").transition().duration(200).delay(200).style("opacity", 1).text("This diagram reveals asymmetries: if a chord is tapered, there are more respondents switching to one brand from the other than vice versa.")
d3.select("#p5").transition().duration(200).delay(200).style("opacity", 1).text("Mouse over a brand name to focus on switching behaviour to and from other brands concerning that specific brand.")
svg.selectAll("path.chord")
.filter(function(d) { return d.source.index != i && d.target.index != i; })
.transition()
.style("stroke-opacity", opacity)
.style("fill-opacity", opacity);
};
}//fadeOut
// Returns an array of tick angles and labels, given a group.
function groupTicks(d) {
var k = (d.endAngle - d.startAngle) / d.value;
return d3.range(0, d.value, 1).map(function(v, i) {
return {
angle: v * k + d.startAngle,
label: i % 5 ? null : v + "%"
};
});
}//groupTicks
</script>
</body>
</html>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment