d3.js Stacked bar chart with Logarithmic Scaling

Question

I am using d3.js to create a stacked bar chart similar to http://mbostock.github.com/d3/ex/stack.html with some additional features.i I was able to add labels and a grid, but I am having some issues implementing a log scale. I am using

d3.scale.log().domain([minNum,maxNum]).range([height,0]) 

but I can not figure out how to implement it with a stacked graph, with the following code:

var vis = d3.select("#chart")
.append("svg")
.attr("width", width)
.attr("height", height + margin);

var layers = vis.selectAll("g.layer")
.data(data)
 .enter().append("g")
.style("fill", function(d, i) { return color(i / (n - 1)); })
.attr("class", "layer");

var bars = layers.selectAll("g.bar")
.data(function(d) { return d; })
.enter().append("g")
.attr("class", "bar")
.attr("transform", function(d) { return "translate(" + x(d) + ",0)"; });

bars.append("rect")
.attr("width", x({x: .9}))
.attr("x", 0)
.attr("y", y1)
.attr("height", function(d) { return y0(d) - y1(d); });

I know it deals with: .attr("height", function(d) { return y0(d) - y1(d); });

Any help would be great.

Answer 1

I ran into this super old thread, here's info on why a log scale almost never makes sense for stacked bars, and rarely for regular bar charts or area charts (ie. charts with shapes that have a base)

Answer 2

I think it would be helpful for you to start out with a basic (non-stacked) bar chart with a logarithmic scale --

You will want to give your scale representation a name, such as:

var myscale = d3.scale.log().domain([minNum,maxNum]).range([height,0]);

Then use this scale later to change to screen space, for example:

.attr("height", function(d) {return myscale(d);})

Here is a basic example for you to look over: http://jsfiddle.net/jsl6906/qAHC2/10/

Answer 3

The short guide.

1. Use stack for the original data set.
2. Use special settings of D3 scale to create a second data set with logarithmic values.
3. Use stack again for the resulting set.

// Original linear data
var linData = [
    [
        { x: 0, y:     0.1 },
        { x: 1, y:     1.0 },
        { x: 2, y:   100.0 },
        { x: 3, y: 10000.0 }
    ],
    [
        { x: 0, y:     0.2 },
        { x: 1, y:     2.0 },
        { x: 2, y:   200.0 },
        { x: 3, y: 20000.0 }
    ],
    [
        { x: 0, y:     0.3 },
        { x: 1, y:     3.0 },
        { x: 2, y:   300.0 },
        { x: 3, y: 30000.0 }
    ]
];


// Calculating y0 for linear data
var stack = d3.layout.stack();
stack(linData);

// Setting conversion values from linear to logarithmic without scaling
var yScaleLog = d3.scale.log()
        .domain([0.01,
            d3.max(linData, function(d) {
                return d3.max(d, function(d) {
                    return d.y0 + d.y;
                });
            })
        ])
        .range([0,
            d3.max(linData, function(d) {
                return d3.max(d, function(d) {
                    return d.y0 + d.y;
                });
            })
        ]);

// Creating new set of data with logarithmized values
var logData = [];
for (var i = 0; i < linData.length; i++) {
    logData[i] = [];
    for (var j = 0; j < linData[i].length; j++) {
        logData[i][j] = {
            x: linData[i][j].x,
            y: yScaleLog(linData[i][j].y)
        };
    }
}

// Calculating y0 for logarithmic data
stack(logData);

Full code: http://jsfiddle.net/romanshamin/5vF6a/2/