I have this dataset for income:
Income Number of people 0 245981 8.8 150444 30 126063 49.9 123519 70 115029 90.7 277149 109.1 355768 130 324246 150.3 353239 170.2 396008 190 396725 210 398640 230.1 401932 250 416079 270 412727 289.8 385192 309.7 343178 329.7 293707 349.6 239982 369.7 201557 389.3 165132 442.3 442075 543.4 196526 679.9 146784 883.9 48600 1555 44644
(As you can see, the width between income levels gets larger towards the end.)
- How do I make an accurate histogram of this data in JavaScript? (On a linear x-axis scale with a range from for example 0 - 2000)
- How do I factor out the number of people to show only percentages at different intervals?
- If I'd like to place exactly 100 symbols representing the data, how do I decide where to place them?
The existing histogram examples are based on computing the histogram from samples, say if you had a list of individual people and their incomes. In this case, you already have the data for the histogram—you just want to display it.
The tricky thing here is that your histogram has variable-width bins. The first thing you can do is ignore the variable-width of each bin and just display a simple lollipop chart. The x-axis is a linear scale for income, and the y-axis is a linear scale for count of people:
http://bl.ocks.org/1624656
If you want to convert this to a histogram, you can't just replace those fixed-width lines with variable-width bars; you need to normalize the data so that the area of the bar encodes the frequency of people with that income. Therefore, the width of the bar is the income range (such as from 0 to 8.8 for the first bin), and the height of the bar is the quantity of people divided by the width. As a result, the area (width × height) is proportional to the number of people. That looks like this:
http://bl.ocks.org/1624660