Hacker News Reader

People think they understand them, but piecharts are hard to understand because you can't differentiate between a 24% and a 25% slice. Read a far better explanation here: http://www.perceptualedge.com/articles/visual_business_intel...

They are perfect for a very, very limited set of datasets: if you have a very large slice and a number of small slices and want to show the obvious difference.

View on HN

Ummm... I think you can make that argument for ANY chart type. I can't differentiate between 25% and 24% in a bar graph either.

In fact, IMO, bar charts (normally) have the far worse problem of being relative to the largest value. In a percentage situation, you want the size of the bars to be relative to the whole, otherwise its misleading to say the least...

So the take away fact is that all visualisations have weaknesses under certain conditions.

But to offer graphing library that does not offer pie charts is, well, a bit silly in my opinion.

As an analyst, I find that the difference between 24% and 25% rarely matters in practice. Pie charts are useful for showing that big differences are big, and that unimportant things are unimportant. Most other uses require a notion of absolute size of the pie in addition to breakdown, which can be communicated in a pie chart, but it's harder.

Reply on news.ycombinator.com