I'm making a bar chart using the Dimensional Charting javascript library dc.js, which is based on d3 and crossfilter.
All I want to do is display a histogram with a specified number of bins, this should be easy using the barChart
function.
I have an array called data
which contains floating-point values between 0 and 90000, and I just want to display the distribution using a histogram with 10 bins.
I use the following code to produce the histogram below:
var cf = crossfilter(data);
var dim = cf.dimension(function(d){ return d[attribute.name]; });
var n_bins = 10;
var xExtent = d3.extent(data, function(d) { return d[attribute.name]; });
var binWidth = (xExtent[1] - xExtent[0]) / n_bins;
grp = dim.group(function(d){return Math.floor(d / binWidth) * binWidth;});
chart = dc.barChart("#" + id_name);
chart.width(200)
.height(180)
.margins({top: 15, right: 10, bottom: 20, left: 40})
.dimension(dim)
.group(grp)
.round(Math.floor)
.centerBar(false)
.x(d3.scale.linear().domain(xExtent).range([0,n_bins]))
.elasticY(true)
.xAxis()
.ticks(4);
That doesn't really look right: each bar is really skinny! I want a normal looking histogram, where the bars are thick and nearly touch each other, with maybe a couple of pixels of padding between each bar. Any idea what I'm doing wrong?
Bar charts in dc.js use the xUnits function to automatically calculate the width of the bars in a histogram based on the range of your x-axis. If you want to set the width to a static value you can use a custom xUnits function for example:
chart.xUnits(function(){return 10;});
This should give you a more fitting width.