I have found that in much of the data that I am looking through, after sorting from largest to smallest, there is a pattern similar to a rotated logistic function. That is, it declines steeply along the vertical axis at first, evens out, and then continues to decline along the vertical axis at the same rate as it declined to start with. Does this tell me anything about the nature of the data underlying the graph outside the obvious? I feel like I might be missing something from this. If this is a ridiculous thought just let me know and I can remove the question.
$\begingroup$
$\endgroup$
1
-
1$\begingroup$ Sounds similar to reliability/survival data, where most of the failure is early or late. $\endgroup$– xanCommented Jun 24, 2014 at 0:26
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
2
If you sort from smallest to largest, plot the value along X, and the index along Y, and the value is the sum accumulated until that point, then you've plotted an empirical cumulative distribution function. http://en.wikipedia.org/wiki/Cumulative_distribution_function
I'll hazard a guess that you plotted the index on the X axis and the value on the Y. This is consistent with the Gaussian/Normal distribution, Cauchy, Student's-t, logistic, etc. distributions - any peaked symmetric distribution will look very similar to the non-discerning eye.
-
$\begingroup$ Thanks Joe, your guess is right. So essentially this is telling me the data is Gaussian? $\endgroup$– 114Commented Jun 24, 2014 at 14:30
-
$\begingroup$ Not necessarily... to establish it's Gaussian, I would use a QQ-plot (en.wikipedia.org/wiki/Q%E2%80%93Q_plot), or use QQ-plots of your data vs. various distributions. $\endgroup$– JoeCommented Jun 24, 2014 at 14:43