×
Namespaces

Variants
Actions
Revision as of 13:40, 10 July 2012 by Oskar Bukolt (Talk | contribs)

HERE Maps API - How to cluster map markers

From Nokia Developer Wiki
Jump to: navigation, search

This article explains how to create clustered markers and reduce cluttering on the map when dealing with a very large number of markers .

Article Metadata
Code ExampleTested with
Devices(s): Internet Explorer, Google Chrome, Firefox
Compatibility
Platform(s): Web
Dependencies: Nokia Maps 2.2.1
Article
Keywords: Nokia Maps, Marker, Cluster, Large Dataset
Created: jasfox (12 Jun 2012)
Last edited: Oskar Bukolt (10 Jul 2012)

Contents

Introduction

When displaying a very large number of markers, it is likely that the map display will look very cluttered. It can be difficult to understand geographical distribution of the given data set, and performance will be reduced as each marker will take time to be rendered. What is required is a method of display the data appropriately at different zooms, showing more details as the user drills down to display a smaller area. Marker clustering achieves this by summarizing the data at higher zoom levels. An illustration of the problem and its solution can be seen below.


UnClusteredMap.png
Map at low zoom displaying 2000 Markers distributed between

Texas and the Eastern Seaboard of the U.S.A.


ClusterMap1.png
Cluster Map at low zoom summarizing 2000 Markers distributed between

Texas and the Eastern Seaboard of the U.S.A.


ClusterMap2.png
Cluster map at a higher zoom, showing the Eastern Seaboard only


Code Commentary

The coded example provided is based upon an Original Library for the Google Maps API. Fortunately, the code is available under an Apache License, which means that it can be freely modified provided that an attribution remains. The Nokia Maps example is a derivative work which has been modified to work with the Nokia Maps API.

When to trigger the cluster function

The code works by making a re-calculation whenever the view port is moved, or the zoom level changed, this can be achieved by adding in a eventListener to the mapviewchangeend Event and an observer to the zoomLevel property,

 var eventHandler = function (evt) {     
me_.resetViewport();
};
 
map_.addListener("mapviewchangeend", eventHandler);
map_.addObserver("zoomLevel", eventHandler);

When the view port is changed, the array of all markers is distributed into two buckets, those withing the current view port and those left outside - Only those which remain on screen are processed further. Each marker left on screen is tested to see if it is close to any markers processed previously. If so it is added to that cluster, otherwise it is added to its own cluster. Essentially the flow of this code is unchanged form the original example. If a cluster has more than one marker, the cluster is rendered using SVG graphics

var markerIcon = createIcon(this.getTotalMarkers());
clusterMarker_ = new nokia.maps.map.Marker(center_ ,
{icon: markerIcon}
);
map_.objects.add ( clusterMarker_);

The SVG graphic consists of three concentric circles with the marker count within it.

'<svg width="39" height="39" xmlns="http://www.w3.org/2000/svg">' +
'<circle stroke="#FFF" fill="__MAINCOLOR__" cx="19" cy="19" r="19" stroke-width="2" />' +
'<circle stroke="#FFF" fill="__MAINCOLOR__" cx="19" cy="19" r="15" stroke-width="2"/>' +
'<circle stroke="#FFF" fill="__MAINCOLOR__" cx="19" cy="19" r="11" stroke-width="2"/>' +
'<text x="19" y="23" font-size="10pt" font-family="arial" font-weight="bold" text-anchor="middle" fill="#FFF" textContent="__TEXTCONTENT__">__TEXT__</text>' +
'</svg>

The color and marker count placeholders are filled in when the image is returned:

var stops = {
"0": "#8A8A8A", //Grey for 0-10
"10": "#CACA00", //Yellow for 11-50
"50": "#00CC00", //Green for 51-100
"100": "#0000CC", //Blue for 101-200
"200": "#CC00CC", //Purple for 201-500
"500": "#FF0000" //Red for over 500
};
var svgParser = new nokia.maps.gfx.SvgParser();
 
createIcon = function (count, mainColor) {
 
[...]
for (var i in stops) {
if (count > i){
color = stops[i];
}
}
 
var svg = iconSVG[digit]
.replace(/__TEXTCONTENT__/g, count)
.replace(/__TEXT__/g, count)
.replace(/__MAINCOLOR__/g, color);
return new nokia.maps.gfx.GraphicsImage(svgParser.parseSvg(svg));
};


Adding an OList wrapper

The code for the addition/removal of the markers for clustering is held in the addMarkers(), addMarker(), clearMarkers(), removeMarker() methods, these are taken from the original library, and have only been converted as necessary. To update these methods to use the familiar nokia.maps.util.OList() interface for adding/removing markers, the original methods have been wrapped as shown:

function MarkerClusterer(map, opt_markers) {
...
ExtendedOList.prototype = new nokia.maps.util.OList();
function ExtendedOList (){
this.addAll = function (elements){
ExtendedOList.prototype.addAll.call(this, elements);
addMarkers(elements);
}
this.add = function (element){
ExtendedOList.prototype.add.call(this, element);
addMarker(element);
}
this.clear= function (){
clearMarkers();
ExtendedOList.prototype.clear.call(this);
}
this.remove= function (element){
removeMarker(element)
ExtendedOList.prototype.remove.call(this);
}
this.removeAt= function (idx){
removeMarker( ExtendedOList.prototype.get.call(this,idx));
ExtendedOList.prototype.removeAt.call(this,idx);
}
 
this.removeAll= function (elements){
for (var i = 0; i < elements.length; ++i) {
removeMarker( elements[i]);
}
ExtendedOList.prototype.removeAll.call(this,elements);
}
}
this.objects = new ExtendedOList();
etc...

How to use the marker clusterer library

The MarkerClusterer library is available as a separate JavaScript file. To use it, include a link to the file in the <header> of your page.

<script type="text/javascript" charset="UTF-8" src="./markerclusterer.js"></script>

To cluster the markers place them into an array, and create the MarkerClusterer as shown, and and call MarkerClusterer.objects.addAll()

var markers = [];
for (var i = 0; i < size; i++){
var marker = new nokia.maps.map.StandardMarker (randomPointNear (lat, long, spread) );
markers.push(marker);
}
 
var markerCluster = new MarkerClusterer();
markerCluster.objects.addAll(markers);

The MarkerClusterer class has been written as a nokia.maps.map.component.Component . To display the clustered markers on a map, attach the component to the MapDisplay.

display.addComponent(markerCluster);

The MarkerClusterer exposes a single objects property for manipulating the markers. As expected this ordered list of objects exposes an nokia.maps.util.OList interface, and markers can be added, removed and queried in the same familar manner as objects in a nokia.maps.map.Container

  • clusterer.objects.add(marker); to add a single marker to the MarkerClusterer
  • clusterer.objects.addAll(markers); to add an Array of markers
  • clusterer.objects.clear(); to empty all markers from the MarkerClusterer
  • clusterer.objects.getLength(); to find the total number of markers clustered.
  • clusterer.objects.remove(marker); to remove a single marker from the MarkerClusterer
  • clusterer.objects.removeAll(markers); to remove an Array of markers from the MarkerClusterer
  • clusterer.objects.removeAt(index); to remove a specified markerat the given index from the MarkerClusterer


Summary

The Code Example adds 7000 random markers to a map and demonstrates the use of the MarkerClusterer on a Nokia Map. This is a client side library which works reasonably well for a large number of markers, but performance is reduced when the number of markers reaches the tens of thousands, where it starts to trigger "slow Script" dialogs in Internet Explorer for example. 'The performance of client side libraries is inextricably linked to the performance of the web browser used. For even larger numbers of markers the only solution would be to trigger the calculations server side instead, to introduce view port based marker management or to introduce a map overlay

513 page views in the last 30 days.
×