centile

You can chart a centile. Each point can be the centile of X measures.

> you can't chart a centile, Why? If you have servers with millions of qps, the graph would show the 99% latency over, say, a second or ten.

Instead these gains have been increasingly captured by the top centile. The share of income by the top 1% has increased over 120% since 1979 [2].

The performance gain is in latency consistency, it seems [0] though this chart is silly, since I'm guessing the X axis is time, and you can't chart a centile, that's ridiculous. .

Or instead of a single number representing the average, show a distribution of the mean total spent by each centile of users. You'll likely see a long flat line at 0 for the first 95% and then a steep curve up for the last 5%.

That's the most valid perspective IMO, but it's going to be useless to the audience, when they can't even understand what a "median" is, much less a centile.

Bin the population into wealth centiles and observe the curves. The Gini coefficient is not a magic arbiter of inequality.

He instead splits the distributions of income into three parts, the bottom 50%, the 40% above that, and the top 10% and simply refers to them by centile and decile. The "top decile" has an unambiguous meaning, whereas "upper class" could mean anything.

I can only access the abstract, but this is fascinating: "[t]he highest centile increased by some 18 to 22 units in the course of the century". An increase of 22 units runs the gamut of possible BMI values [1], from severely underweight to obese class 2.

Below the household name corporations are so many in real need of serious minds, and I know personally that the salaries are not as bad as they are made to look at with statistics when debated here, I mean you think a typical character here is going to be a bottom decile performer, or top centile producer, compared against the big wobbly world of corporate computer departments? Edit horrid auto correction

The notion of giving some kind of "performance bonus" to mothers for the mathematical talent of their children is absurd, since they have no impact on any innate talent whatsoever; I dread to think what kind of an impact this would have when the proposed bonus is high enough to put them in the 99th centile of income.

For these, the sample mean, the sample variance, the sample centiles -- they all severely underestimate the true values. What's worse is when these tools start to remove "outliers". With a heavy-tailed distribution, the majority of samples don't contribute very much at all to the expectation. The strongest signal is found in the extreme values. The strongest signal is found in the stuff that is thrown out. The junk that's left is the noise, the stuff that doesn't tell you very much about what you're dealing with. I stand firm in my belief that unless you can prove how CLT applies to your input distributions, you should not assume normality. And if you don't know what you are doing, stop reporting means. Stop reporting centiles.