freebsd-dev

History

Ed Maste a9cf54b0c9 ministat: disallow negative variance / nan Stddev With all values identical it was possible for Var() to return a negative value due to limited floating point precision, resulting in "nan" reported as Stddev. Variance cannot actually be negative, so just return 0. We can later investigate alternate algorithms for calculating variance to reduce the effect of catastrophic cancellation here. Reported by: Arshan Khanifar <arshankhanifar_gmail.com> Approved by: phk Sponsored by: The FreeBSD Foundation		2018-02-21 15:54:23 +00:00
..
chameleon	A small statistics tool for gauging the statistical significance	2003-08-13 07:21:54 +00:00
iguana	A small statistics tool for gauging the statistical significance	2003-08-13 07:21:54 +00:00
Makefile	Convert to usr.bin/ to LIBADD	2014-11-25 14:29:10 +00:00
Makefile.depend	DIRDEPS_BUILD: Update dependencies.	2017-10-31 00:07:04 +00:00
ministat.1	Clarify the ministat default width	2015-03-26 17:13:11 +00:00
ministat.c	ministat: disallow negative variance / nan Stddev	2018-02-21 15:54:23 +00:00
README

README

$FreeBSD$

A small tool to do the statistics legwork on benchmarks etc.

Prepare your data into two files, one number per line
run 
	./ministat data_before data_after

and see what it says.

You need at least three data points in each data set, but the more
you have the better your result generally gets.

Here are two typical outputs:

x _1
+ _2
+--------------------------------------------------------------------------+
|x            +    x+      x            x   x             +           ++   |
|        |_________|______AM_______________|__A___________M_______________||
+--------------------------------------------------------------------------+
    N           Min           Max        Median           Avg        Stddev
x   5         36060         36138         36107       36105.6     31.165686
+   5         36084         36187         36163       36142.6     49.952978
No difference proven at 95.0% confidence

Here nothing can be concluded from the numbers.  It _may_ be possible to
prove something if many more measurements are made, but with only five
measurements, nothing is proven.


x _1
+ _2
+--------------------------------------------------------------------------+
|                                                               +          |
|                               x                               +         +|
|x                    x         x          x                    +         +|
|         |_______________A_____M_________|                   |_M___A____| |
+--------------------------------------------------------------------------+
    N           Min           Max        Median           Avg        Stddev
x   5         0.133         0.137         0.136        0.1354  0.0015165751
+   5         0.139          0.14         0.139        0.1394 0.00054772256
Difference at 95.0% confidence
        0.004 +/- 0.00166288
        2.95421% +/- 1.22812%
        (Student's t, pooled s = 0.00114018)

Here we have a clearcut difference, not very big, but clear and unambiguous.