13 KiB

Raw Blame History

Bar Graph

In this section, we'll use AJAX to build a bar graph

Lesson Objectives

Use AJAX to make an asynchronous call to an external data file
Create a Bar graph

Set up

Let's create our standard setup in index.html:

<!DOCTYPE html>
<html lang="en" dir="ltr">
    <head>
        <link rel="stylesheet" href="app.css">
    </head>
    <body>
        <svg></svg>
        <script src="https://d3js.org/d3.v5.min.js"></script>
        <script src="app.js" charset="utf-8"></script>
    </body>
</html>

app.js:

var WIDTH = 800;
var HEIGHT = 600;

d3.select('svg')
    .style('width', WIDTH)
    .style('height', HEIGHT);

app.css:

svg {
    border:1px solid black;
}

This is what we should have:

Create an external file to hold our data

Let's create an external data.json file, which will hold fake data regarding how often job posts require certain skills

[
  {
    "name": "HTML",
    "count": 21
  },
  {
    "name": "CSS",
    "count": 17
  },
  {
    "name": "Responsive Web Design",
    "count": 17
  },
  {
    "name": "JavaScript",
    "count": 17
  },
  {
    "name": "Git",
    "count": 16
  },
  {
    "name": "Angular.js",
    "count": 9
  },
  {
    "name": "Node.js",
    "count": 9
  },
  {
    "name": "PostgreSQL",
    "count": 8
  },
  {
    "name": "Agile Project Management",
    "count": 8
  },
  {
    "name": "MongoDB",
    "count": 7
  },
  {
    "name": "Trello",
    "count": 7
  },
  {
    "name": "Testing / TDD",
    "count": 7
  },
  {
    "name": "jQuery",
    "count": 7
  },
  {
    "name": "User Testing",
    "count": 6
  },
  {
    "name": "MySQL",
    "count": 6
  },
  {
    "name": "PHP",
    "count": 6
  },
  {
    "name": "React.js",
    "count": 6
  },
  {
    "name": "AJAX",
    "count": 6
  },
  {
    "name": "Express.js",
    "count": 5
  },
  {
    "name": "Heroku",
    "count": 5
  },
  {
    "name": "Wireframing",
    "count": 5
  },
  {
    "name": "Sass/SCSS",
    "count": 5
  },
  {
    "name": "Mobile Web",
    "count": 4
  },
  {
    "name": "Rails",
    "count": 4
  },
  {
    "name": "WordPress",
    "count": 4
  },
  {
    "name": "Drupal",
    "count": 3
  },
  {
    "name": "Ruby",
    "count": 3
  },
  {
    "name": "Ember.js",
    "count": 3
  },
  {
    "name": "Python",
    "count": 3
  },
  {
    "name": "Amazon EC2",
    "count": 2
  },
  {
    "name": "Computer Science degree",
    "count": 1
  },
  {
    "name": "Backbone.js",
    "count": 1
  },
  {
    "name": "Less",
    "count": 1
  },
  {
    "name": "Prototyping",
    "count": 1
  },
  {
    "name": "Redis",
    "count": 1
  }
]

Make an AJAX Request

Write the basic code:

D3 has lots of different methods for making AJAX requests to file of different data types:

d3.json('path').then(function(data){
    //do something with the json data here
});
d3.csv('path').then(function(data){
    //do something with the csv data here
});
d3.tsv('path').then(function(data){
    //do something with the tsv data here
});
d3.xml('path').then(function(data){
    //do something with the xml data here
});
d3.html('path').then(function(data){
    //do something with the html data here
});
d3.text('path').then(function(data){
    //do something with the text data here
});

Since our data is in JSON format, we'll use the first kind of call:

var WIDTH = 800;
var HEIGHT = 600;

d3.select('svg')
    .style('width', WIDTH)
    .style('height', HEIGHT);

d3.json('data.json').then(function(data){
    console.log(data);
});

Handle file access

If you opened the index.html file in the browser directly, instead of serving it on a web server, you'll notice we've encountered an error. Check your developer console:

The issue here is that web browsers are not supposed to make AJAX requests to files on your computer. If it could, this would be a huge security flaw. Let's create a basic file server. To do this, you'll need to install Node.js. Once that's done, open up your computer's terminal

Mac: command + space then type terminal and hit enter
Windows: click Start, type cmd and hit enter

Next type the following into your terminal:

npm install -g http-server

If you get error messages try

sudo npm install -g http-server

This installs a basic http server that was built using Node.js. To run it, use the terminal to navigate to the directory where you code is (type cd to change folders in the terminal) and run the following:

http-server .

You should see something like this:

Now go to http://localhost:8080/ in your browser. You should now see that your AJAX call is succeeding (if you have issues, hold down shift and hit the refresh button to force the browser to reload all files that may have been cached):

Use AJAX data to create SVG elements

Now that our AJAX calls are succeeding, let's start building our app. From here on out, it's all basic JavaScript and D3. Note that everything we'll write for the rest of this lesson is done within the success callback of our AJAX request. In production we might want to move this code elsewhere, but for now this is easier for learning. Let's create some rectangles for our bar graph:

d3.json('data.json').then(function(data){
    d3.select('svg').selectAll('rect')
        .data(data)
        .enter()
        .append('rect');
});

Our Elements tab in our dev tools should look something like this:

Adjust the height/width of the bars

Let's create a scale that maps the count property of each element in data to a visual height for the corresponding bar. We'll use a linear scale. Remember to map the HEIGHT of the graph to a very low data point and the top of the graph (0 in the range) map to a very high data value:

var yScale = d3.scaleLinear();
yScale.range([HEIGHT, 0]);
var yMin = d3.min(data, function(datum, index){
    return datum.count;
})
var yMax = d3.max(data, function(datum, index){
    return datum.count;
})
yScale.domain([yMin, yMax]);

We could use d3.extent, but we're going to need the individual min/max values later on. Immediatley after the above code, let's tell D3 to adjust the height using the yScale. Remember that the Y axis is flipped. A low data value produces a high range value. But a even though the range is high, the bar itself should be small. We'll need to re-flip the values just for height so that a low data value produces a small bar and a high data value produces a large bar. To do this, let's subtract whatever the range point is from the HEIGHT of the graph. This way, if yScale(datum.count) produces, say, 500, the height of the bar will be 100. We can use yScale(datum.count) normally when adjusting the position of the bars later.

d3.selectAll('rect')
    .attr('height', function(datum, index){
        return HEIGHT-yScale(datum.count);
    });

Now our rectangles have height, but no width:

In our app.css let's give all our bars the same width:

rect {
    width: 15px;
}

Adjust the horizontal/vertical placement of the bars

Our bars all overlap each other at the moment. Let's space them out by mapping x position to index in the data array:

var xScale = d3.scaleLinear();
xScale.range([0, WIDTH]);
xScale.domain([0, data.length]);
d3.selectAll('rect')
    .attr('x', function(datum, index){
        return xScale(index);
    });

This maps indices in the in the array to horizontal range points

Now let's move the bars so they grow from the bottom, not the hang from the top. Now a high data point produces a low range value which doesn't push a large bow down much. A low data point produces a high range value which pushes a small bar down a lot.

d3.selectAll('rect')
    .attr('y', function(datum, index){
        return yScale(datum.count);
    });

Our last few bars don't have any height, because we've mapped the minimum count property of our data to a visual range value of 0 in our yScale. Let's adjust this code:

var yScale = d3.scaleLinear();
yScale.range([HEIGHT, 0]);
var yMin = d3.min(data, function(datum, index){
    return datum.count;
})
var yMax = d3.max(data, function(datum, index){
    return datum.count;
})
yScale.domain([yMin, yMax]);

to be this code:

var yScale = d3.scaleLinear();
yScale.range([HEIGHT, 0]);
var yMin = d3.min(data, function(datum, index){
    return datum.count;
})
var yMax = d3.max(data, function(datum, index){
    return datum.count;
})
yScale.domain([yMin-1, yMax]); //adjust this line

Now the domain min is 1 less than what's actually in our data set. Domains with the original min are treated as higher values than what's expected for the min of the graph. We get this:

Make width of bars dynamic

Currently, our bars have a fixed width. No matter how many elements we have, they have 15px width. If we had more data elements, the bars could overlap. Let's change this. Since each rect will be the same width, no matter what the data is, we can just assign width a computed value:

d3.selectAll('rect')
    .attr('width', WIDTH/data.length);

Now let's adjust our rect css so our bars are more visible:

rect {
    /*  remove the width rule that was here */
    stroke:white;
    stroke-width:1px;
}

Change the color of the bar based on data

Right now the bars are black. A linear scale will interpolate between colors just like a regular number:

var yDomain = d3.extent(data, function(datum, index){
    return datum.count;
})
var colorScale = d3.scaleLinear();
colorScale.domain(yDomain)
colorScale.range(['#00cc00', 'blue'])
d3.selectAll('rect')
    .attr('fill', function(datum, index){
        return colorScale(datum.count)
    })

We'll need to recalculate the yDomain so that the real min of the data set is used to map #00cc00:

Add axes

The left axis is just like before:

var leftAxis = d3.axisLeft(yScale);
d3.select('svg')
    .append('g').attr('id', 'left-axis')
    .call(leftAxis);

To create the bottom axis, we need to be able to map strings to points on a domain. We'll use a band scale for this, which just divides up the range into equal parts and maps it to an array of discrete values (values that can't be interpolated. e.g. strings):

var skillScale = d3.scaleBand();
var skillDomain = data.map(function(skill){
    return skill.name
});
skillScale.range([0, WIDTH]);
skillScale.domain(skillDomain);

Notice we use data.map(). This is regular javascript which simply loops through an array and modifies each element based on the given function. It then returns the resulting array. In the above example, skillDomain will be an array containing the various name properties of each of the data elements.

Once we have an array of each of the skills, we use this as the domain and map each skill to a point within the range. Remember the point in the range is created by dividing up the full range equally based on the number of elements in the domain.

Now that we have a scale which maps each skill text to a point in the x range, we can create the bottom axis as before:

var bottomAxis = d3.axisBottom(skillScale);
d3.select('svg')
    .append('g').attr('id', 'bottom-axis')
    .call(bottomAxis)
    .attr('transform', 'translate(0,'+HEIGHT+')');

We still need to stop the <svg> from clipping everything:

svg {
    overflow: visible;
}

Our result:

The text is all cluttered, though. Let's use some CSS to fix this:

#bottom-axis text {
    transform:rotate(45deg);
}

It's rotated, but it's rotated around the center of the element. Let's change this, so it rotates around the start of the text:

#bottom-axis text {
    transform:rotate(45deg);
    text-anchor: start;
}

Let's move the graph to the right, so we can see the values for the left axis:

svg {
    overflow: visible;
    margin-left: 20px;
}

Conclusion

In this chapter we learned how to use AJAX to make an asynchronous request that will populate a bar graph. In the next chapter we'll create a pie chart that animates when you remove sections from it.

13 KiB Raw Blame History