Cumulative standard deviation: Difference between revisions
Content deleted Content added
PatGarrett (talk | contribs) |
m →{{header|Wren}}: Minor tidy |
||
(195 intermediate revisions by 65 users not shown) | |||
Line 1:
{{task|Probability and statistics}}
{{task heading}}
Write a stateful function, class, generator or co-routine that takes a series of floating point numbers, ''one at a time'', and returns the running [[wp:Standard Deviation|standard deviation]] of the series.
The task implementation should use the most natural programming style of those listed for the function in the implementation language; the task ''must'' state which is being used.
Do not apply [[wp:Bessel's correction|Bessel's correction]]; the returned standard deviation should always be computed as if the sample seen so far is the entire population.
;Test case:
Use this to compute the standard deviation of this demonstration set, <math>\{2, 4, 4, 4, 5, 5, 7, 9\}</math>, which is <math>2</math>.
;Related tasks:
* [[Random numbers]]
{{Related tasks/Statistical measures}}
<hr>
=={{header|11l}}==
{{trans|Python:_Callable_class}}
<syntaxhighlight lang="11l">T SD
sum = 0.0
sum2 = 0.0
n = 0.0
F ()(x)
.sum += x
.sum2 += x ^ 2
.n += 1.0
R sqrt(.sum2 / .n - (.sum / .n) ^ 2)
V sd_inst = SD()
L(value) [2, 4, 4, 4, 5, 5, 7, 9]
print(value‘ ’sd_inst(value))</syntaxhighlight>
{{out}}
<pre>
2 0
4 1
4 0.942809042
4 0.866025404
5 0.979795897
5 1
7 1.399708424
9 2
</pre>
=={{header|360 Assembly}}==
For maximum compatibility, this program uses only the basic instruction set.
Part of the code length is due to the square root algorithm and to the nice output.
<syntaxhighlight lang="360asm">******** Standard deviation of a population
STDDEV CSECT
USING STDDEV,R13
SAVEAREA B STM-SAVEAREA(R15)
DC 17F'0'
DC CL8'STDDEV'
STM STM R14,R12,12(R13)
ST R13,4(R15)
ST R15,8(R13)
LR R13,R15
SR R8,R8 s=0
SR R9,R9 ss=0
SR R4,R4 i=0
LA R6,1
LH R7,N
LOOPI BXH R4,R6,ENDLOOPI
LR R1,R4 i
BCTR R1,0
SLA R1,1
LH R5,T(R1)
ST R5,WW ww=t(i)
MH R5,=H'1000' w=ww*1000
AR R8,R5 s=s+w
LR R15,R5
MR R14,R5 w*w
AR R9,R15 ss=ss+w*w
LR R14,R8 s
SRDA R14,32
DR R14,R4 /i
ST R15,AVG avg=s/i
LR R14,R9 ss
SRDA R14,32
DR R14,R4 ss/i
LR R2,R15 ss/i
LR R15,R8 s
MR R14,R8 s*s
LR R3,R15
LR R15,R4 i
MR R14,R4 i*i
LR R1,R15
LA R14,0
LR R15,R3
DR R14,R1 (s*s)/(i*i)
SR R2,R15
LR R10,R2 std=ss/i-(s*s)/(i*i)
LR R11,R10 std
SRA R11,1 x=std/2
LR R12,R10 px=std
LOOPWHIL EQU *
CR R12,R11 while px<>=x
BE ENDWHILE
LR R12,R11 px=x
LR R15,R10 std
LA R14,0
DR R14,R12 /px
LR R1,R12 px
AR R1,R15 px+std/px
SRA R1,1 /2
LR R11,R1 x=(px+std/px)/2
B LOOPWHIL
ENDWHILE EQU *
LR R10,R11
CVD R4,P8 i
MVC C17,MASK17
ED C17,P8
MVC BUF+2(1),C17+15
L R1,WW
CVD R1,P8
MVC C17,MASK17
ED C17,P8
MVC BUF+10(1),C17+15
L R1,AVG
CVD R1,P8
MVC C18,MASK18
ED C18,P8
MVC BUF+17(5),C18+12
CVD R10,P8 std
MVC C18,MASK18
ED C18,P8
MVC BUF+31(5),C18+12
WTO MF=(E,WTOMSG)
B LOOPI
ENDLOOPI EQU *
L R13,4(0,R13)
LM R14,R12,12(R13)
XR R15,R15
BR R14
DS 0D
N DC H'8'
T DC H'2',H'4',H'4',H'4',H'5',H'5',H'7',H'9'
WW DS F
AVG DS F
P8 DS PL8
MASK17 DC C' ',13X'20',X'2120',C'-'
MASK18 DC C' ',10X'20',X'2120',C'.',3X'20',C'-'
C17 DS CL17
C18 DS CL18
WTOMSG DS 0F
DC H'80',XL2'0000'
BUF DC CL80'N=1 ITEM=1 AVG=1.234 STDDEV=1.234 '
YREGS
END STDDEV</syntaxhighlight>
{{out}}
<pre>N=1 ITEM=2 AVG=2.000 STDDEV=0.000
N=2 ITEM=4 AVG=3.000 STDDEV=1.000
N=3 ITEM=4 AVG=3.333 STDDEV=0.942
N=4 ITEM=4 AVG=3.500 STDDEV=0.866
N=5 ITEM=5 AVG=3.800 STDDEV=0.979
N=6 ITEM=5 AVG=4.000 STDDEV=1.000
N=7 ITEM=7 AVG=4.428 STDDEV=1.399
N=8 ITEM=9 AVG=5.000 STDDEV=2.000</pre>
=={{header|Action!}}==
{{libheader|Action! Tool Kit}}
{{libheader|Action! Real Math}}
<syntaxhighlight lang="action!">INCLUDE "H6:REALMATH.ACT"
REAL sum,sum2
INT count
PROC Calc(REAL POINTER x,sd)
REAL tmp1,tmp2,tmp3
RealAdd(sum,x,tmp1) ;tmp1=sum+x
RealAssign(tmp1,sum) ;sum=sum+x
RealMult(x,x,tmp1) ;tmp1=x*x
RealAdd(sum2,tmp1,tmp2) ;tmp2=sum2+x*x
RealAssign(tmp2,sum2) ;sum2=sum2+x*x
count==+1
IF count=0 THEN
IntToReal(0,sd) ;sd=0
ELSE
IntToReal(count,tmp1)
RealMult(sum,sum,tmp2) ;tmp2=sum*sum
RealDiv(tmp2,tmp1,tmp3) ;tmp3=sum*sum/count
RealDiv(tmp3,tmp1,tmp2) ;tmp2=sum*sum/count/count
RealDiv(sum2,tmp1,tmp3) ;tmp3=sum2/count
RealSub(tmp3,tmp2,tmp1) ;tmp1=sum2/count-sum*sum/count/count
Sqrt(tmp1,sd) ;sd=sqrt(sum2/count-sum*sum/count/count)
FI
RETURN
PROC Main()
INT ARRAY values=[2 4 4 4 5 5 7 9]
INT i
REAL x,sd
Put(125) PutE() ;clear screen
MathInit()
IntToReal(0,sum)
IntToReal(0,sum2)
count=0
FOR i=0 TO 7
DO
IntToReal(values(i),x)
Calc(x,sd)
Print("x=") PrintR(x)
Print(" sum=") PrintR(sum)
Print(" sd=") PrintRE(sd)
OD
RETURN</syntaxhighlight>
{{out}}
[https://gitlab.com/amarok8bit/action-rosetta-code/-/raw/master/images/Cumulative_standard_deviation.png Screenshot from Atari 8-bit computer]
<pre>
x=2 sum=2 sd=0
x=4 sum=6 sd=1
x=4 sum=10 sd=.942809052
x=4 sum=14 sd=.86602541
x=5 sum=19 sd=.979795903
x=5 sum=24 sd=1
x=7 sum=31 sd=1.39970843
x=9 sum=40 sd=1.99999999
</pre>
=={{header|Ada}}==
<syntaxhighlight lang="ada">
with Ada.Numerics.Elementary_Functions; use Ada.Numerics.Elementary_Functions;
with Ada.Numerics.Elementary_Functions; use Ada.Numerics.Elementary_Functions;
with Ada.Text_IO; use Ada.Text_IO;
with Ada.Float_Text_IO; use Ada.Float_Text_IO;
with Ada.Integer_Text_IO; use Ada.Integer_Text_IO;
procedure Test_Deviation is
type Sample is record
N : Natural := 0;
end record;
procedure Add (Data : in out Sample; Point : Float) is
begin
Data.N := Data.N + 1;
Data.
Data.
end Add;
function Deviation (Data : Sample) return Float is
begin
return Sqrt (Data.
end Deviation;
Data : Sample;
Test : array (1..8) of
begin
for
Add (Data, Float(Test
Put("N="); Put(Item => Index, Width => 1);
Put(" ITEM="); Put(Item => Test(Index), Width => 1);
Put(" AVG="); Put(Item => Float(Data.Sum)/Float(Index), Fore => 1, Aft => 3, Exp => 0);
Put(" STDDEV="); Put(Item => Deviation (Data), Fore => 1, Aft => 3, Exp => 0);
New_line;
end loop;
end Test_Deviation;
</syntaxhighlight>
{{out}}
<pre>
N=1 ITEM=2 AVG=2.000 STDDEV=0.000
N=2 ITEM=4 AVG=3.000 STDDEV=1.000
N=3 ITEM=4 AVG=3.333 STDDEV=0.943
N=4 ITEM=4 AVG=3.500 STDDEV=0.866
N=5 ITEM=5 AVG=3.800 STDDEV=0.980
N=6 ITEM=5 AVG=4.000 STDDEV=1.000
N=7 ITEM=7 AVG=4.429 STDDEV=1.400
N=8 ITEM=9 AVG=5.000 STDDEV=2.000
</pre>
=={{header|ALGOL 68}}==
{{trans|C}}
Line 43 ⟶ 283:
{{works with|ALGOL 68|Standard - no extensions to language used}}
{{works with|ALGOL 68G|Any - tested with release
<!-- {{works with|ELLA ALGOL 68|Any (with appropriate job cards) - tested with release [http://sourceforge.net/projects/algol68/files/algol68toc/algol68toc-1.8.8d/algol68toc-1.8-8d.fc9.i386.rpm/download 1.8.8d.fc9.i386]}} -->
Note: the use of a UNION to mimic C's enumerated types is "experimental" and probably not typical of "production code". However it is a example of '''ALGOL 68'''s ''conformity CASE clause'' useful for classroom dissection.
<
STDDEV = STRUCT(CHAR stddev),
MEAN = STRUCT(CHAR mean),
Line 53 ⟶ 293:
COUNT = STRUCT(CHAR count),
RESET = STRUCT(CHAR reset);
MODE ACTION = UNION ( VALUE, STDDEV, MEAN, VAR, COUNT, RESET );
LONG REAL sum := 0;
LONG REAL sum2 := 0;
INT num := 0;
PROC stat object = (LONG REAL v, ACTION action)LONG REAL:
(
Line 86 ⟶ 326:
ESAC
);
[]LONG REAL v = ( 2,4,4,4,5,5,7,9 );
Line 94 ⟶ 334:
FOR i FROM LWB v TO UPB v DO
sd := stat object(v[i], LOC VALUE);
printf(($"value: "g(0,6)," standard dev := "g(0,6)l$, v[i], sd))
OD
)</syntaxhighlight>
{{out}}
<pre>
value: 2.000000 standard dev :=
value: 4.000000 standard dev := 1.000000
value: 4.000000 standard dev := .942809
value: 4.000000 standard dev := .866025
value: 5.000000 standard dev := .979796
value: 5.000000 standard dev := 1.000000
value: 7.000000 standard dev := 1.399708
value: 9.000000 standard dev := 2.000000
</pre>
{{trans|python}}
{{works with|ALGOL 68|Standard - no extensions to language used}}
{{works with|ALGOL 68G|Any - tested with release 2.8-win32}}
<!-- {{works with|ELLA ALGOL 68|Any (with appropriate job cards) - tested with release [http://sourceforge.net/projects/algol68/files/algol68toc/algol68toc-1.8.8d/algol68toc-1.8-8d.fc9.i386.rpm/download 1.8.8d.fc9.i386]}} -->
A code sample in an object oriented style:
<
LONG REAL sum,
LONG REAL sum2,
INT num
);
OP INIT = (REF STAT new)REF STAT:
(init OF class stat)(new);
MODE CLASSSTAT = STRUCT(
PROC (REF STAT, LONG REAL #value#)VOID plusab,
Line 126 ⟶ 373:
CLASSSTAT class stat;
plusab OF class stat := (REF STAT self, LONG REAL value)VOID:(
num OF self +:= 1;
Line 132 ⟶ 379:
sum2 OF self +:= value*value
);
OP +:= = (REF STAT lhs, LONG REAL rhs)VOID: # some syntatic sugar #
(plusab OF class stat)(lhs, rhs);
stddev OF class stat := (REF STAT self)LONG REAL:
long sqrt((variance OF class stat)(self));
OP STDDEV = ([]LONG REAL value)LONG REAL: ( # more syntatic sugar #
REF STAT stat = INIT LOC STAT;
Line 146 ⟶ 393:
(stddev OF class stat)(stat)
);
mean OF class stat := (REF STAT self)LONG REAL:
sum OF self/LONG REAL(num OF self);
variance OF class stat := (REF STAT self)LONG REAL:(
LONG REAL m = (mean OF class stat)(self);
sum2 OF self/LONG REAL(num OF self)-m*m
);
count OF class stat := (REF STAT self)LONG REAL:
num OF self;
init OF class stat := (REF STAT self)REF STAT:(
sum OF self := sum2 OF self := num OF self := 0;
self
);
[]LONG REAL value = ( 2,4,4,4,5,5,7,9 );
main:
(
# printf(($"standard deviation operator = "g(0,6)l$, STDDEV value));
#
REF STAT stat = INIT LOC STAT;
FOR i FROM LWB value TO UPB value DO
stat +:= value[i];
printf(($"value: "g(0,6)," standard dev := "g(0,6)l$, value[i], (stddev OF class stat)(stat)))
OD
#
;
printf(($"standard deviation = "g(0,6)l$, (stddev OF class stat)(stat)));
printf(($"mean = "g(0,6)l$, (mean OF class stat)(stat)));
printf(($"variance = "g(0,6)l$, (variance OF class stat)(stat)));
printf(($"count = "g(0,6)l$, (count OF class stat)(stat)))
#
)
</syntaxhighlight>
{{out}}
<pre>
value: 4.000000 standard
value: 4.000000 standard dev := .942809
value: 5.000000 standard dev := .979796
value: 5.000000 standard dev := 1.000000
value: 7.000000 standard dev := 1.399708
value: 9.000000 standard dev := 2.000000
</pre>
{{trans|python}}
{{works with|ALGOL 68|Standard - no extensions to language used}}
{{works with|ALGOL 68G|Any - tested with release [http://sourceforge.net/projects/algol68/files/algol68g/algol68g-1.18.0/algol68g-1.18.0-9h.tiny.el5.centos.fc11.i386.rpm/download 1.18.0-9h.tiny]}}
<!-- {{works with|ELLA ALGOL 68|Any (with appropriate job cards) - tested with release [http://sourceforge.net/projects/algol68/files/algol68toc/algol68toc-1.8.8d/algol68toc-1.8-8d.fc9.i386.rpm/download 1.8.8d.fc9.i386]}} -->
A simple - but "unpackaged" - code example, useful if the standard deviation is required on only one set of concurrent data:
<
INT n;
Line 210 ⟶ 466:
LONG REAL value = values[i];
printf(($2(xg(0,6))l$, value, sd(value)))
OD</syntaxhighlight>
{{out}}
<pre>
2.000000 .000000
Line 221 ⟶ 478:
9.000000 2.000000
</pre>
=={{header|ALGOL W}}==
{{Trans|ALGOL 68}}
This is an Algol W version of the third, "unpackaged" Algol 68 sample, which was itself translated from Python.
<syntaxhighlight lang="algolw">begin
long real sum, sum2;
integer n;
long real procedure sd (long real value x) ;
begin
sum := sum + x;
sum2 := sum2 + (x*x);
n := n + 1;
if n = 0 then 0 else longsqrt(sum2/n - sum*sum/n/n)
end sd;
sum := sum2 := n := 0;
r_format := "A"; r_w := 14; r_d := 6; % set output to fixed point format %
for i := 2,4,4,4,5,5,7,9
do begin
long real val;
val := i;
write(val, sd(val))
end for_i
end.</syntaxhighlight>
{{out}}
<pre>
2.000000 0.000000
4.000000 1.000000
4.000000 0.942809
4.000000 0.866025
5.000000 0.979795
5.000000 1.000000
7.000000 1.399708
9.000000 2.000000
</pre>
=={{header|AppleScript}}==
Accumulation over a fold:
<syntaxhighlight lang="applescript">-------------- CUMULATIVE STANDARD DEVIATION -------------
-- stdDevInc :: Accumulator -> Num -> Index -> Accumulator
-- stdDevInc :: {sum:, squaresSum:, stages:} -> Real -> Integer
-- -> {sum:, squaresSum:, stages:}
on stdDevInc(a, n, i)
set sum to (sum of a) + n
set squaresSum to (squaresSum of a) + (n ^ 2)
set stages to (stages of a) & ¬
((squaresSum / i) - ((sum / i) ^ 2)) ^ 0.5
{sum:(sum of a) + n, squaresSum:squaresSum, stages:stages}
end stdDevInc
--------------------------- TEST -------------------------
on run
set xs to [2, 4, 4, 4, 5, 5, 7, 9]
stages of foldl(stdDevInc, ¬
{sum:0, squaresSum:0, stages:[]}, xs)
--> {0.0, 1.0, 0.942809041582, 0.866025403784, 0.979795897113, 1.0, 1.399708424448, 2.0}
end run
-------------------- GENERIC FUNCTIONS -------------------
-- foldl :: (a -> b -> a) -> a -> [b] -> a
on foldl(f, startValue, xs)
tell mReturn(f)
set v to startValue
set lng to length of xs
repeat with i from 1 to lng
set v to |λ|(v, item i of xs, i, xs)
end repeat
return v
end tell
end foldl
-- mReturn :: First-class m => (a -> b) -> m (a -> b)
on mReturn(f)
-- 2nd class handler function lifted into 1st class script wrapper.
if script is class of f then
f
else
script
property |λ| : f
end script
end if
end mReturn</syntaxhighlight>
{{Out}}
<syntaxhighlight lang="applescrip">{0.0, 1.0, 0.942809041582, 0.866025403784,
0.979795897113, 1.0, 1.399708424448, 2.0}</syntaxhighlight>
Or as a map-accumulation:
<syntaxhighlight lang="applescript">-------------- CUMULATIVE STANDARD DEVIATION -------------
-- cumulativeStdDevns :: [Float] -> [Float]
on cumulativeStdDevns(xs)
script go
on |λ|(sq, x, i)
set {s, q} to sq
set _s to x + s
set _q to q + (x ^ 2)
{{_s, _q}, ((_q / i) - ((_s / i) ^ 2)) ^ 0.5}
end |λ|
end script
item 2 of mapAccumL(go, {0, 0}, xs)
end cumulativeStdDevns
--------------------------- TEST -------------------------
on run
cumulativeStdDevns({2, 4, 4, 4, 5, 5, 7, 9})
end run
------------------------- GENERIC ------------------------
-- foldl :: (a -> b -> a) -> a -> [b] -> a
on foldl(f, startValue, xs)
tell mReturn(f)
set v to startValue
set lng to length of xs
repeat with i from 1 to lng
set v to |λ|(v, item i of xs, i, xs)
end repeat
return v
end tell
end foldl
-- mapAccumL :: (acc -> x -> (acc, y)) -> acc -> [x] -> (acc, [y])
on mapAccumL(f, acc, xs)
-- 'The mapAccumL function behaves like a combination of map and foldl;
-- it applies a function to each element of a list, passing an
-- accumulating parameter from |Left| to |Right|, and returning a final
-- value of this accumulator together with the new list.' (see Hoogle)
script
on |λ|(a, x, i)
tell mReturn(f) to set pair to |λ|(item 1 of a, x, i)
{item 1 of pair, (item 2 of a) & {item 2 of pair}}
end |λ|
end script
foldl(result, {acc, []}, xs)
end mapAccumL
-- mReturn :: First-class m => (a -> b) -> m (a -> b)
on mReturn(f)
-- 2nd class handler function lifted into 1st class script wrapper.
if script is class of f then
f
else
script
property |λ| : f
end script
end if
end mReturn</syntaxhighlight>
{{Out}}
<pre>{0.0, 1.0, 0.942809041582, 0.866025403784, 0.979795897113, 1.0, 1.399708424448, 2.0}</pre>
=={{header|Arturo}}==
<syntaxhighlight lang="rebol">arr: new []
loop [2 4 4 4 5 5 7 9] 'value [
'arr ++ value
print [value "->" deviation arr]
]</syntaxhighlight>
{{out}}
<pre>2 -> 0.0
4 -> 1.0
4 -> 0.9428090415820634
4 -> 0.8660254037844386
5 -> 0.9797958971132711
5 -> 0.9999999999999999
7 -> 1.39970842444753
9 -> 2.0</pre>
=={{header|AutoHotkey}}==
<syntaxhighlight lang="autohotkey">Data := [2,4,4,4,5,5,7,9]
for k, v in Data {
FileAppend, % "#" a_index " value = " v " stddev = " stddev(v) "`n", * ; send to stdout
}
return
stddev(x) {
static n, sum, sum2
n++
sum += x
sum2 += x*x
return sqrt((sum2/n) - (((sum*sum)/n)/n))
}</syntaxhighlight>
{{out}}
<pre>
#1 value = 2 stddev 0 0.000000
#2 value = 4 stddev 0 1.000000
#3 value = 4 stddev 0 0.942809
#4 value = 4 stddev 0 0.866025
#5 value = 5 stddev 0 0.979796
#6 value = 5 stddev 0 1.000000
#7 value = 7 stddev 0 1.399708
#8 value = 9 stddev 0 2.000000
</pre>
=={{header|AWK}}==
<syntaxhighlight lang="awk">
# syntax: GAWK -f STANDARD_DEVIATION.AWK
BEGIN {
Line 256 ⟶ 725:
return(sqrt(variance))
}
</syntaxhighlight>
{{out}}
<pre>
2 0
Line 268 ⟶ 737:
9 2
</pre>
=={{header|Axiom}}==
{{incorrect|Axiom|It does not return the ''running'' standard deviation of the series.}}
We implement a domain with dependent type T with the operation + and identity 0:<syntaxhighlight lang="axiom">)abbrev package TESTD TestDomain
TestDomain(T : Join(Field,RadicalCategory)): Exports == Implementation where
R ==> Record(n : Integer, sum : T, ssq : T)
Line 285 ⟶ 756:
sd obj ==
mean : T := obj.sum / (obj.n::T)
sqrt(obj.ssq / (obj.n::T) - mean*mean)</
D ==> TestDomain(T)
items := [2,4,4,4,5,5,7,9+x] :: List T;
Line 298 ⟶ 769:
(2) 2
Type: Expression(Integer)</
=={{header|BBC BASIC}}==
{{works with|BBC BASIC for Windows}}
Uses the MOD(array()) and SUM(array()) functions.
<
FOR i% = 1 TO 8
READ n
Line 317 ⟶ 788:
i% += 1
list(i%) = n
= SQR(MOD(list())^2/i% - (SUM(list())/i%)^2)</
{{out}}
<pre>
Value = 2, running SD = 0
Line 332 ⟶ 803:
=={{header|C}}==
<
#include <stdlib.h>
#include <math.h>
Line 379 ⟶ 850:
so->sum2 += v*v;
return stat_obj_value(so, so->action);
}</
<
int main()
Line 393 ⟶ 864:
FREE_STAT_OBJECT(so);
return 0;
}</
=={{header|C sharp|C#}}==
<
using System.Collections.Generic;
using System.Linq;
Line 446 ⟶ 891:
}
}
}</
<pre>0
1
Line 456 ⟶ 901:
2</pre>
=={{header|
No attempt to handle different types -- standard deviation is intrinsically a real number.
<syntaxhighlight lang="cpp">
#include <cassert>
#include <cmath>
#include <vector>
#include <iostream>
template<int N> struct MomentsAccumulator_
{
std::vector<double> m_;
MomentsAccumulator_() : m_(N + 1, 0.0) {}
void operator()(double v)
{
double inc = 1.0;
for (auto& mi : m_)
{
mi += inc;
inc *= v;
}
}
};
double Stdev(const std::vector<double>& moments)
{
assert(moments.size() > 2);
assert(moments[0] > 0.0);
const double mean = moments[1] / moments[0];
const double meanSquare = moments[2] / moments[0];
return sqrt(meanSquare - mean * mean);
}
int main(void)
{
std::vector<int> data({ 2, 4, 4, 4, 5, 5, 7, 9 });
MomentsAccumulator_<2> accum;
for (auto d : data)
{
accum(d);
std::cout << "Running stdev: " << Stdev(accum.m_) << "\n";
}
}
</syntaxhighlight>
=={{header|Clojure}}==
<syntaxhighlight lang="lisp">
(defn stateful-std-deviation[x]
(letfn [(std-dev[x]
(let [v (deref (find-var (symbol (str *ns* "/v"))))]
(swap! v conj x)
(let [m (/ (reduce + @v) (count @v))]
(Math/sqrt (/ (reduce + (map #(* (- m %) (- m %)) @v)) (count @v))))))]
(when (nil? (resolve 'v))
(intern *ns* 'v (atom [])))
(std-dev x)))
</syntaxhighlight>
=={{header|COBOL}}==
{{works with|OpenCOBOL|1.1}}
<syntaxhighlight lang="cobol">IDENTIFICATION DIVISION.
PROGRAM-ID. run-stddev.
environment division.
input-output section.
file-control.
select input-file assign to "input.txt"
organization is line sequential.
data division.
file section.
fd input-file.
01 inp-record.
03 inp-fld pic 9(03).
working-storage section.
01 filler pic 9(01) value 0.
88 no-more-input value 1.
01 ws-tb-data.
03 ws-tb-size pic 9(03).
03 ws-tb-table.
05 ws-tb-fld pic s9(05)v9999 comp-3 occurs 0 to 100 times
depending on ws-tb-size.
01 ws-stddev pic s9(05)v9999 comp-3.
PROCEDURE DIVISION.
move 0 to ws-tb-size
open input input-file
read input-file
at end
set no-more-input to true
end-read
perform
until no-more-input
add 1 to ws-tb-size
move inp-fld to ws-tb-fld (ws-tb-size)
call 'stddev' using by reference ws-tb-data
ws-stddev
display 'inp=' inp-fld ' stddev=' ws-stddev
read input-file at end set no-more-input to true end-read
end-perform
close input-file
stop run.
end program run-stddev.
IDENTIFICATION DIVISION.
data division.
working-storage section.
01 ws-tbx pic s9(03) comp.
01 ws-tb-work.
03 ws-sum pic s9(05)v9999 comp-3 value +0.
03 ws-sumsq pic s9(05)v9999 comp-3 value +0.
linkage section.
01 ws-tb-data.
03 ws-tb-size pic 9(03).
03 ws-tb-table.
05 ws-tb-fld pic s9(05)v9999 comp-3 occurs 0 to 100 times
depending on ws-tb-size.
01 ws-stddev pic s9(05)v9999 comp-3.
PROCEDURE DIVISION using ws-tb-data ws-stddev.
compute ws-sum = 0
perform test before varying ws-tbx from 1 by +1 until ws-tbx > ws-tb-size
compute ws-sum = ws-sum + ws-tb-fld (ws-tbx)
end-perform
compute ws-avg rounded = ws-sum / ws-tb-size
compute ws-sumsq = 0
perform test before varying ws-tbx from 1 by +1 until ws-tbx > ws-tb-size
compute ws-sumsq = ws-sumsq
+ (ws-tb-fld (ws-tbx) - ws-avg) ** 2.0
end-perform
compute ws-stddev = ( ws-sumsq / ws-tb-size) ** 0.5
goback.
end program stddev.
</syntaxhighlight>
<syntaxhighlight lang="cobol">sample output:
inp=002 stddev=+00000.0000
inp=004 stddev=+00001.0000
inp=004 stddev=+00000.9427
inp=004 stddev=+00000.8660
inp=005 stddev=+00000.9797
inp=005 stddev=+00001.0000
inp=007 stddev=+00001.3996
inp=009 stddev=+00002.0000
</syntaxhighlight>
=={{header|CoffeeScript}}==
Uses a class instance to maintain state.
<
class StandardDeviation
constructor: ->
Line 567 ⟶ 1,076:
"""
</syntaxhighlight>
{{out}}
<pre>
Values: 2
Line 597 ⟶ 1,106:
=={{header|Common Lisp}}==
Since we don't care about the sample values once std dev is computed, we only need to keep track of their sum and square sums, hence:<syntaxhighlight lang="lisp">(defun running-stddev ()
(let ((sum 0) (sq 0) (n 0))
(lambda (x)
Line 644 ⟶ 1,112:
(/ (sqrt (- (* n sq) (* sum sum))) n))))
CL-USER> (loop with f = (running-stddev) for i in '(2 4 4 4 5 5 7 9) do
(format t "~a ~a~%" i (funcall f i)))
NIL
2 0.0
4 1.0
4 0.94280905
4 0.8660254
5 0.97979593
5 1.0
7 1.3997085
9 2.0</syntaxhighlight>
In the REPL, one step at a time:
<syntaxhighlight lang="lisp">CL-USER> (setf fn (running-stddev))
#<Interpreted Closure (:INTERNAL RUNNING-STDDEV) @ #x21b9a492>
CL-USER> (funcall fn 2)
0.0
CL-USER> (funcall fn 4)
1.0
CL-USER> (funcall fn 4)
0.94280905
CL-USER> (funcall fn 4)
0.8660254
CL-USER> (funcall fn 5)
0.97979593
CL-USER> (funcall fn 5)
1.0
CL-USER> (funcall fn 7)
1.3997085
CL-USER> (funcall fn 9)
2.0
</syntaxhighlight>
=={{header|Component Pascal}}==
{{incorrect|Component Pascal|Function does not take numbers individually.}}
BlackBox Component Builder
<
MODULE StandardDeviation;
IMPORT StdLog, Args,Strings,Math;
Line 693 ⟶ 1,192:
END Do;
END StandardDeviation.
</syntaxhighlight>
Execute: ^Q StandardDeviation.Do 2 4 4 4 5 5 7 9 ~<br/>
{{out}}
<pre>
1 :> 0.0
Line 706 ⟶ 1,205:
8 :> 2.0
</pre>
=={{header|Crystal}}==
===Object===
Use an object to keep state.
{{trans|Ruby}}
<syntaxhighlight lang="ruby">class StdDevAccumulator
def initialize
@n, @sum, @sum2 = 0, 0.0, 0.0
end
def <<(num)
@n += 1
@sum += num
@sum2 += num**2
Math.sqrt (@sum2 * @n - @sum**2) / @n**2
end
end
sd = StdDevAccumulator.new
i = 0
[2,4,4,4,5,5,7,9].each { |n| puts "adding #{n}: stddev of #{i+=1} samples is #{sd << n}" }</syntaxhighlight>
{{out}}
<pre>
adding 2: stddev of 1 samples is 0.0
adding 4: stddev of 2 samples is 1.0
adding 4: stddev of 3 samples is 0.9428090415820634
adding 4: stddev of 4 samples is 0.8660254037844386
adding 5: stddev of 5 samples is 0.9797958971132712
adding 5: stddev of 6 samples is 1.0
adding 7: stddev of 7 samples is 1.3997084244475304
adding 9: stddev of 8 samples is 2.0
</pre>
===Closure===
{{trans|Ruby}}
<syntaxhighlight lang="ruby">def sdaccum
n, sum, sum2 = 0, 0.0, 0.0
->(num : Int32) do
n += 1
sum += num
sum2 += num**2
Math.sqrt( (sum2 * n - sum**2) / n**2 )
end
end
sd = sdaccum
[2,4,4,4,5,5,7,9].each {|n| print sd.call(n), ", "}</syntaxhighlight>
{{out}}
<pre>0.0, 1.0, 0.9428090415820634, 0.8660254037844386, 0.9797958971132712, 1.0, 1.3997084244475304, 2.0</pre>
=={{header|D}}==
<
struct StdDev {
Line 734 ⟶ 1,283:
writefln("%e", stdev.getStdDev());
}
}</
{{out}}
<pre>0.000000e+00
1.000000e+00
Line 746 ⟶ 1,295:
=={{header|Delphi}}==
See: [[#Pascal]]
=={{header|E}}==
This implementation produces two (function) objects sharing state.
It is idiomatic in E to separate input from output (read from write)
rather than combining them into one object.
The algorithm is {{trans|Perl}} and the results were checked against [[#Python]].
<
var sum := 0.0
var sumSquares := 0.0
Line 790 ⟶ 1,328:
return [insert, stddev]
}</
<
# value: <insert>, <stddev>
Line 809 ⟶ 1,347:
1.0
1.3997084244475297
2.0</
=={{header|EasyLang}}==
{{trans|Pascal}}
<syntaxhighlight lang="easylang">
global sum sum2 n .
proc sd x . r .
sum += x
sum2 += x * x
n += 1
r = sqrt (sum2 / n - sum * sum / n / n)
.
v[] = [ 2 4 4 4 5 5 7 9 ]
for v in v[]
sd v r
print v & " " & r
.
</syntaxhighlight>
=={{header|Elixir}}==
{{trans|Erlang}}
<syntaxhighlight lang="elixir">defmodule Standard_deviation do
def add_sample( pid, n ), do: send( pid, {:add, n} )
def create, do: spawn_link( fn -> loop( [] ) end )
def destroy( pid ), do: send( pid, :stop )
def get( pid ) do
send( pid, {:get, self()} )
receive do
{ :get, value, _pid } -> value
end
end
def task do
pid = create()
for x <- [2,4,4,4,5,5,7,9], do: add_print( pid, x, add_sample(pid, x) )
destroy( pid )
end
defp add_print( pid, n, _add ) do
IO.puts "Standard deviation #{ get(pid) } when adding #{ n }"
end
defp loop( ns ) do
receive do
{:add, n} -> loop( [n | ns] )
{:get, pid} ->
send( pid, {:get, loop_calculate( ns ), self()} )
loop( ns )
:stop -> :ok
end
end
defp loop_calculate( ns ) do
average = loop_calculate_average( ns )
:math.sqrt( loop_calculate_average( for x <- ns, do: :math.pow(x - average, 2) ) )
end
defp loop_calculate_average( ns ), do: Enum.sum( ns ) / length( ns )
end
Standard_deviation.task</syntaxhighlight>
{{out}}
<pre>
Standard deviation 0.0 when adding 2
Standard deviation 1.0 when adding 4
Standard deviation 0.9428090415820634 when adding 4
Standard deviation 0.8660254037844386 when adding 4
Standard deviation 0.9797958971132712 when adding 5
Standard deviation 1.0 when adding 5
Standard deviation 1.3997084244475302 when adding 7
Standard deviation 2.0 when adding 9
</pre>
=={{header|Emacs Lisp}}==
<syntaxhighlight lang="lisp">(defun running-std (items)
(let ((running-sum 0)
(running-len 0)
(running-squared-sum 0)
(result 0))
(dolist (item items)
(setq running-sum (+ running-sum item))
(setq running-len (1+ running-len))
(setq running-squared-sum (+ running-squared-sum (* item item)))
(setq result (sqrt (- (/ running-squared-sum (float running-len))
(/ (* running-sum running-sum)
(float (* running-len running-len))))))
(message "%f" result))
result))
{{out}}
1.
0.942809
0.866025
0.979796
1.
1.399708
2.000000
2.0
{{libheader|Calc}}
<syntaxhighlight lang="lisp">(let ((x '(2 4 4 4 5 5 7 9)))
(string-to-number (calc-eval "sqrt(vpvar($1))" nil (append '(vec) x))))</syntaxhighlight>
{{libheader|generator.el}}
<syntaxhighlight lang="lisp">;; lexical-binding: t
(require 'generator)
(iter-defun std-dev-gen (lst)
(let ((sum 0)
(avg 0)
(tmp '())
(std 0))
(dolist (i lst)
(setq i (float i))
(push i tmp)
(setq sum (+ sum i))
(setq avg (/ sum (length tmp)))
(setq std 0)
(dolist (j tmp)
(setq std (+ std (expt (- j avg) 2))))
(setq std (/ std (length tmp)))
(setq std (sqrt std))
(iter-yield std))))
(let* ((test-data '(2 4 4 4 5 5 7 9))
(generator (std-dev-gen test-data)))
(dolist (i test-data)
(message "with %d: %f" i (iter-next generator))))</syntaxhighlight>
=={{header|Erlang}}==
<syntaxhighlight lang="erlang">
-module( standard_deviation ).
Line 874 ⟶ 1,511:
[add_print(Pid, X, add_sample(Pid, X)) || X <- [2,4,4,4,5,5,7,9]],
destroy( Pid ).
add_print( Pid, N, _Add ) -> io:fwrite( "Standard deviation ~p when adding ~p~n", [get(Pid), N] ).
Line 893 ⟶ 1,528:
loop_calculate_average( Ns ) -> lists:sum( Ns ) / erlang:length( Ns ).
</syntaxhighlight>
{{out}}
<pre>
Line 908 ⟶ 1,543:
=={{header|Factor}}==
<
sequences ;
IN: standard-deviator
Line 929 ⟶ 1,564:
{ 2 4 4 4 5 5 7 9 }
<standard-deviator> [ [ add-value ] curry each ] keep
current-std number>string print ;</
=={{header|FOCAL}}==
<syntaxhighlight lang="focal">01.01 C-- TEST SET
01.10 S T(1)=2;S T(2)=4;S T(3)=4;S T(4)=4
01.20 S T(5)=5;S T(6)=5;S T(7)=7;S T(8)=9
01.30 D 2.1
01.35 T %6.40
01.40 F I=1,8;S A=T(I);D 2.2;T "VAL",A;D 2.3;T " SD",A,!
01.50 Q
02.01 C-- RUNNING STDDEV
02.02 C-- 2.1: INITIALIZE
02.03 C-- 2.2: INSERT VALUE A
02.04 C-- 2.3: A = CURRENT STDDEV
02.10 S XN=0;S XS=0;S XQ=0
02.20 S XN=XN+1;S XS=XS+A;S XQ=XQ+A*A
02.30 S A=FSQT(XQ/XN - (XS/XN)^2)</syntaxhighlight>
{{out}}
<pre>VAL= 2.00000 SD= 0.00000
VAL= 4.00000 SD= 1.00000
VAL= 4.00000 SD= 0.94281
VAL= 4.00000 SD= 0.86603
VAL= 5.00000 SD= 0.97980
VAL= 5.00000 SD= 1.00000
VAL= 7.00000 SD= 1.39971
VAL= 9.00000 SD= 2.00000</pre>
=={{header|Forth}}==
<
: st-count ( stats -- n ) f@ ;
: st-sum ( stats -- sum ) float+ f@ ;
: st-sumsq ( stats -- sum*sum ) 2 floats + f@ ;
: st-mean ( stats -- mean )
Line 952 ⟶ 1,611:
: st-stddev ( stats -- stddev )
st-variance fsqrt ;
: st-add ( fnum stats -- )
dup
1e dup f+! float+
fdup dup f+! float+
fdup f* f+!
std-stddev ;</syntaxhighlight>
This variation is more numerically stable when there are large numbers of samples or large sample ranges.
<
: st-mean ( stats -- mean ) float+ f@ ;
: st-nvar ( stats -- n*var ) 2 floats + f@ ;
Line 962 ⟶ 1,629:
: st-add ( x stats -- )
dup
1e dup f+! \ update count
fdup dup st-mean f- fswap
Line 969 ⟶ 1,637:
float+ dup f+! \ update mean
( delta x )
dup f@ f- f* float+ f+!
st-stddev ;</syntaxhighlight>
Usage example:
<syntaxhighlight lang="forth">create stats 0e f, 0e f, 0e f,
2e stats st-add f. \ 0.
4e stats st-add f. \ 1.
4e stats st-add f. \ 0.942809041582063
4e stats st-add f. \ 0.866025403784439
5e stats st-add f. \ 0.979795897113271
5e stats st-add f. \ 1.
7e stats st-add f. \ 1.39970842444753
9e stats st-add f. \ 2.
</syntaxhighlight>
=={{header|Fortran}}==
{{works with|Fortran|2003 and later}}
<syntaxhighlight lang="fortran">
program standard_deviation
implicit none
integer(kind=4), parameter :: dp = kind(0.0d0)
real(kind=dp), dimension(:), allocatable :: vals
integer(kind=4) :: i
real(kind=dp), dimension(8) :: sample_data = (/ 2, 4, 4, 4, 5, 5, 7, 9 /)
do i = lbound(sample_data, 1), ubound(sample_data, 1)
call sample_add(vals, sample_data(i))
write(*, fmt='(''#'',I1,1X,''value = '',F3.1,1X,''stddev ='',1X,F10.8)') &
i, sample_data(i), stddev(vals)
end do
if (allocated(vals)) deallocate(vals)
contains
! Adds value :val: to array :population: dynamically resizing array
subroutine sample_add(population, val)
real(kind=dp), dimension(:), allocatable, intent (inout) :: population
real(kind=dp), intent (in) :: val
real(kind=dp), dimension(:), allocatable :: tmp
integer(kind=4) :: n
if (.not. allocated(population)) then
allocate(population(1))
population(1) = val
else
n = size(population)
call move_alloc(population, tmp)
allocate(population(n + 1))
population(1:n) = tmp
population(n + 1) = val
endif
end subroutine sample_add
! Calculates standard deviation for given set of values
real(kind=dp), dimension(:), intent(in) ::
integer(kind=4) :: n
n = size(vals)
mean = sum(vals)/n
stddev = sqrt(sum((vals - mean)**2)/n)
end function stddev
end program standard_deviation
</syntaxhighlight>
{{out}}
<pre>
#1 value = 2.0 stddev = 0.00000000
#2 value = 4.0 stddev = 1.00000000
#3 value = 4.0 stddev = 0.94280904
#4 value = 4.0 stddev = 0.86602540
#5 value = 5.0 stddev = 0.97979590
#6 value = 5.0 stddev = 1.00000000
#7 value = 7.0 stddev = 1.39970842
#8 value = 9.0 stddev = 2.00000000
</pre>
===Old style, four ways===
Early computers loaded the entire programme and its working storage into memory and left it there throughout the run. Uninitialised variables would start with whatever had been left in memory at their address by whatever last used those addresses, though some systems would clear all of memory to zero or possibly some other value before each load. Either way, if a routine was invoked a second time, its variables would have the values left in them by their previous invocation. The DATA statement allows initial values to be specified, and allows repeat counts when specifying such values as well. It is not an executable statement: it is not re-executed on second and subsequent invocations of the containing routine. Thus, it is easy to have a routine employ counters and the like, visible only within themselves and initialised to zero or whatever suited.
With more complex operating systems, routines that relied on retaining values across invocations might no longer work - perhaps a fresh version of the routine would be loaded to memory (perhaps at odd intervals), or, on exit, the working storage would be discarded. There was a half-way scheme, whereby variables that had appeared in DATA statements would be retained while the others would be discarded. This subtle indication has been discarded in favour of the explicit SAVE statement, naming those variables whose values are to be retained between invocations, though compilers might also offer an option such as "automatic" (for each invocation always allocate then discard working memory) and "static" (retain values), possibly introducing non-standard keywords as well. Otherwise, the routines would have to use storage global to them such as additional parameters, or, COMMON storage and in later Fortran, the MODULE arrangements for shared items. The persistence of such storage can still be limited, but by naming them in the main line can be ensured for the life of the run. The other routines with access to such storage could enable re-initialisation, additional reports, or multiple accumulations, etc.
Since the standard deviation can be calculated in a single pass through the data, producing values for the standard deviation of all values so far supplied is easily done without re-calculation. Accuracy is quite another matter. Calculations using deviances from a working mean are much better, and capturing the first X as the working mean would be easy, just test on N = 0. The sum and sum-of-squares method is quite capable of generating a negative variance, but the second method cannot, because the terms being added in to V are never negative. This is demonstrated by comparing the results computed from StdDev(A), StdDev(A + 10), StdDev(A + 100), StdDev(A + 1000), etc.
Incidentally, Fortran implementations rarely enable re-entrancy for the WRITE statement, so, since here the functions are invoked in a WRITE statement, the functions cannot themselves use WRITE statements, say for debugging.
<syntaxhighlight lang="fortran">
REAL FUNCTION STDDEV(X) !Standard deviation for successive values.
REAL X !The latest value.
REAL V !Scratchpad.
INTEGER N !Ongoing: count of the values.
REAL EX,EX2 !Ongoing: sum of X and X**2.
SAVE N,EX,EX2 !Retain values from one invocation to the next.
DATA N,EX,EX2/0,0.0,0.0/ !Initial values.
N = N + 1 !Another value arrives.
EX = X + EX !Augment the total.
EX2 = X**2 + EX2 !Augment the sum of squares.
V = EX2/N - (EX/N)**2 !The variance, but, it might come out negative!
STDDEV = SIGN(SQRT(ABS(V)),V) !Protect the SQRT, but produce a negative result if so.
END FUNCTION STDDEV !For the sequence of received X values.
REAL FUNCTION STDDEVP(X) !Standard deviation for successive values.
REAL X !The latest value.
INTEGER N !Ongoing: count of the values.
REAL A,V !Ongoing: average, and sum of squared deviations.
SAVE N,A,V !Retain values from one invocation to the next.
DATA N,A,V/0,0.0,0.0/ !Initial values.
N = N + 1 !Another value arrives.
V = (N - 1)*(X - A)**2 /N + V !First, as it requires the existing average.
A = (X - A)/N + A != [x + (n - 1).A)]/n: recover the total from the average.
STDDEVP = SQRT(V/N) !V can never be negative, even with limited precision.
END FUNCTION STDDEVP !For the sequence of received X values.
REAL FUNCTION STDDEVW(X) !Standard deviation for successive values.
REAL X !The latest value.
REAL V,D !Scratchpads.
INTEGER N !Ongoing: count of the values.
REAL EX,EX2 !Ongoing: sum of X and X**2.
REAL W !Ongoing: working mean.
SAVE N,EX,EX2,W !Retain values from one invocation to the next.
DATA N,EX,EX2/0,0.0,0.0/ !Initial values.
IF (N.LE.0) W = X !Take the first value as the working mean.
N = N + 1 !Another value arrives.
D = X - W !Its deviation from the working mean.
EX = D + EX !Augment the total.
EX2 = D**2 + EX2 !Augment the sum of squares.
V = EX2/N - (EX/N)**2 !The variance, but, it might come out negative!
STDDEVW = SIGN(SQRT(ABS(V)),V) !Protect the SQRT, but produce a negative result if so.
END FUNCTION STDDEVW !For the sequence of received X values.
REAL FUNCTION STDDEVPW(X) !Standard deviation for successive values.
REAL X !The latest value.
INTEGER N !Ongoing: count of the values.
REAL A,V !Ongoing: average, and sum of squared deviations.
REAL W !Ongoing: working mean.
SAVE N,A,V,W !Retain values from one invocation to the next.
DATA N,A,V/0,0.0,0.0/ !Initial values.
IF (N.LE.0) W = X !Oh for self-modifying code!
N = N + 1 !Another value arrives.
D = X - W !Its deviation from the working mean.
V = (N - 1)*(D - A)**2 /N + V !First, as it requires the existing average.
A = (D - A)/N + A != [x + (n - 1).A)]/n: recover the total from the average.
STDDEVPW = SQRT(V/N) !V can never be negative, even with limited precision.
END FUNCTION STDDEVPW !For the sequence of received X values.
PROGRAM TEST
INTEGER I !A stepper.
REAL A(8) !The example data.
DATA A/2.0,3*4.0,2*5.0,7.0,9.0/ !Alas, another opportunity to use @ passed over.
REAL B !An offsetting base.
WRITE (6,1)
1 FORMAT ("Progressive calculation of the standard deviation."/
1 " I",7X,"A(I) EX EX2 Av V*N Ed Ed2 wAv V*N")
B = 1000000 !Provoke truncation error.
DO I = 1,8 !Step along the data series,
WRITE (6,2) I,INT(A(I) + B), !No fractional part, so I don't want F11.0.
1 STDDEV(A(I) + B),STDDEVP(A(I) + B), !Showing progressive values.
2 STDDEVW(A(I) + B),STDDEVPW(A(I) + B) !These with a working mean.
2 FORMAT (I2,I11,1X,4F12.6) !Should do for the example.
END DO !On to the next value.
END
</syntaxhighlight>
Output: the second pair of columns have the calculations done with a working mean and thus accumulate deviations from that.
Progressive calculation of the standard deviation.
I A(I) EX EX2 Av V*N Ed Ed2 wAv V*N
1 2 0.000000 0.000000 0.000000 0.000000
2 4 1.000000 1.000000 1.000000 1.000000
3 4 0.942809 0.942809 0.942809 0.942809
4 4 0.866025 0.866025 0.866025 0.866025
5 5 0.979796 0.979796 0.979796 0.979796
6 5 1.000000 1.000000 1.000000 1.000000
7 7 1.399708 1.399708 1.399708 1.399708
8 9 2.000000 2.000000 2.000000 2.000000
I A(I) EX EX2 Av V*N Ed Ed2 wAv V*N
1 12 0.000000 0.000000 0.000000 0.000000
2 14 1.000000 1.000000 1.000000 1.000000
3 14 0.942809 0.942809 0.942809 0.942809
4 14 0.866025 0.866025 0.866025 0.866025
5 15 0.979796 0.979796 0.979796 0.979796
6 15 1.000000 1.000000 1.000000 1.000000
7 17 1.399708 1.399708 1.399708 1.399708
8 19 2.000000 2.000000 2.000000 2.000000
I A(I) EX EX2 Av V*N Ed Ed2 wAv V*N
1 102 0.000000 0.000000 0.000000 0.000000
2 104 1.000000 1.000000 1.000000 1.000000
3 104 0.942809 0.942809 0.942809 0.942809
4 104 0.866025 0.866025 0.866025 0.866025
5 105 0.979796 0.979796 0.979796 0.979796
6 105 1.000000 0.999999 1.000000 1.000000
7 107 1.399708 1.399708 1.399708 1.399708
8 109 2.000000 1.999999 2.000000 2.000000
I A(I) EX EX2 Av V*N Ed Ed2 wAv V*N
1 1002 0.000000 0.000000 0.000000 0.000000
2 1004 1.000000 1.000000 1.000000 1.000000
3 1004 0.942809 0.942809 0.942809 0.942809
4 1004 0.866025 0.866028 0.866025 0.866025
5 1005 0.979796 0.979798 0.979796 0.979796
6 1005 1.000000 1.000004 1.000000 1.000000
7 1007 1.399708 1.399711 1.399708 1.399708
8 1009 2.000000 1.999997 2.000000 2.000000
I A(I) EX EX2 Av V*N Ed Ed2 wAv V*N
1 10002 -2.000000 0.000000 0.000000 0.000000
2 10004 -1.000000 1.000000 1.000000 1.000000
3 10004 -0.666667 0.942809 0.942809 0.942809
4 10004 1.936492 0.866072 0.866025 0.866025
5 10005 2.181742 0.979829 0.979796 0.979796
6 10005 2.309401 1.000060 1.000000 1.000000
7 10007 1.801360 1.399745 1.399708 1.399708
8 10009 2.645751 1.999987 2.000000 2.000000
I A(I) EX EX2 Av V*N Ed Ed2 wAv V*N
1 100002 19.493589 0.000000 0.000000 0.000000
2 100004 7.416198 1.000000 1.000000 1.000000
3 100004 -7.333333 0.942809 0.942809 0.942809
4 100004 20.093531 0.865650 0.866025 0.866025
5 100005 -1.280625 0.979531 0.979796 0.979796
6 100005 -16.492422 1.000305 1.000000 1.000000
7 100007 17.851427 1.399895 1.399708 1.399708
8 100009 20.566963 1.999835 2.000000 2.000000
I A(I) EX EX2 Av V*N Ed Ed2 wAv V*N
1 1000002 -80.024994 0.000000 0.000000 0.000000
2 1000004 158.767120 1.000000 1.000000 1.000000
3 1000004 -89.146576 0.942809 0.942809 0.942809
4 1000004 90.795097 0.869074 0.866025 0.866025
5 1000005 193.357590 0.981953 0.979796 0.979796
6 1000005 238.361069 0.999691 1.000000 1.000000
7 1000007 153.462296 1.399519 1.399708 1.399708
8 1000009 151.284500 1.997653 2.000000 2.000000
Speaking loosely, to square a number of d digits accurately requires the ability to represent 2d digits accurately, with more digits needed if many such squares are to be added together accurately. In this example, 1000 when squared, is pushing at the limits of single precision. The average&variance method is resistant to this problem (and does not generate negative variances either!) because it manipulates differences from the running average, but it is still better to use a working mean.
In other words, a two-pass method will be more accurate (where the second pass calculates the variance by accumulating deviations from the actual average, itself calculated with a working mean) but at the cost of that second pass and the saving of all the values. Higher precision variables for the accumulations are the easiest way towards accurate results.
=={{header|FreeBASIC}}==
<syntaxhighlight lang="freebasic">' FB 1.05.0 Win64
Function calcStandardDeviation(number As Double) As Double
Static a() As Double
Redim Preserve a(0 To UBound(a) + 1)
Dim ub As UInteger = UBound(a)
a(ub) = number
Dim sum As Double = 0.0
For i As UInteger = 0 To ub
sum += a(i)
Next
Dim mean As Double = sum / (ub + 1)
Dim diff As Double
sum = 0.0
For i As UInteger = 0 To ub
diff = a(i) - mean
sum += diff * diff
Next
Return Sqr(sum/ (ub + 1))
End Function
Dim a(0 To 7) As Double = {2, 4, 4, 4, 5, 5, 7, 9}
For i As UInteger = 0 To 7
Print "Added"; a(i); " SD now : "; calcStandardDeviation(a(i))
Next
Print
Print "Press any key to quit"
Sleep</syntaxhighlight>
{{out}}
<pre>
Added 2 SD now : 0
Added 4 SD now : 1
Added 4 SD now : 0.9428090415820634
Added 4 SD now : 0.8660254037844386
Added 5 SD now : 0.9797958971132712
Added 5 SD now : 1
Added 7 SD now : 1.39970842444753
Added 9 SD now : 2
</pre>
=={{header|Go}}==
Algorithm to reduce rounding errors from WP article.
State maintained with a closure.
<
import (
Line 1,113 ⟶ 1,948:
fmt.Println(r(x))
}
}</
{{out}}
<pre>
0
Line 1,128 ⟶ 1,963:
=={{header|Groovy}}==
Solution:
<
def
samples << sample
def sum = samples.sum()
def sumSq = samples.sum { it * it }
def count = samples.size()
(sumSq/count - (sum/count)**2)**0.5
}
[2,4,4,4,5,5,7,9].each {
}</syntaxhighlight>
{{out}}
<pre>
1
1
=={{header|Haskell}}==
Line 1,152 ⟶ 1,991:
We store the state in the <code>ST</code> monad using an <code>STRef</code>.
<syntaxhighlight lang="haskell">{-# LANGUAGE BangPatterns #-}
import Data.List (foldl') -- '
import Data.STRef
import Control.Monad.ST
sumLen :: [Double] -> Pair Double Double
sumLen = fiof2 . foldl' (\(Pair s l) x -> Pair (s+x) (l+1)) (Pair 0.0 0) --'
where fiof2 (Pair s l) = Pair s (fromIntegral l)
divl :: Pair Double Double -> Double
divl (Pair _ 0.0) = 0.0
divl (Pair s l) = s / l
sd :: [Double] -> Double
sd xs = sqrt $ foldl' (\a x -> a+(x-m)^2) 0 xs / l --'
where p@(Pair s l) = sumLen xs
m = divl p
mkSD :: ST s (Double -> ST s Double)
mkSD = go <$> newSTRef []
where go acc x =
modifySTRef acc (x:) >> (sd <$> readSTRef acc)
main = mapM_ print $ runST $
mkSD >>= forM [2.0, 4.0, 4.0, 4.0, 5.0, 5.0, 7.0, 9.0]</syntaxhighlight>
Or, perhaps more simply, as a map-accumulation over an indexed list:
<syntaxhighlight lang="haskell">import Data.List (mapAccumL)
-------------- CUMULATIVE STANDARD DEVIATION -------------
cumulativeStdDevns :: [Float] -> [Float]
cumulativeStdDevns = snd . mapAccumL go (0, 0) . zip [1.0..]
where
go (s, q) (i, x) =
let _s = s + x
_q = q + (x ^ 2)
in ((_s, _q), sqrt ((_q / i) - ((_s / i) ^ 2)))
--------------------------- TEST -------------------------
main :: IO ()
main = mapM_ print $ cumulativeStdDevns [2, 4, 4, 4, 5, 5, 7, 9]</syntaxhighlight>
{{Out}}
<pre>0.0
1.0
0.9428093
0.8660254
0.97979593
1.0
1.3997087
2.0</pre>
=={{header|Haxe}}==
<
class Main {
Line 1,197 ⟶ 2,074:
return Math.sqrt(average(store));
}
}</
<pre>0
1
Line 1,208 ⟶ 2,085:
=={{header|HicEst}}==
<
set = (2,4,4,4,5,5,7,9)
Line 1,223 ⟶ 2,100:
sum2 = sum2 + x*x
stdev = ( sum2/k - (sum/k)^2) ^ 0.5
END</
<pre>Adding 2 stdev = 0
Adding 4 stdev = 1
Line 1,234 ⟶ 2,111:
=={{header|Icon}} and {{header|Unicon}}==
<
stddev() # reset state / empty
Line 1,242 ⟶ 2,119:
end
procedure stddev(x)
static X,sumX,sum2X
Line 1,255 ⟶ 2,132:
return sqrt( (sum2X / *X) - (sumX / *X)^2 )
}
end</
{{out}}
<pre>stddev (so far) := 0.0
stddev (so far) := 1.0
stddev (so far) := 0.9428090415820626
Line 1,264 ⟶ 2,142:
stddev (so far) := 1.39970842444753
stddev (so far) := 2.0</pre>
=={{header|IS-BASIC}}==
<syntaxhighlight lang="is-basic">100 PROGRAM "StDev.bas"
110 LET N=8
120 NUMERIC ARR(1 TO N)
130 FOR I=1 TO N
140 READ ARR(I)
150 NEXT
160 DEF STDEV(N)
170 LET S1,S2=0
180 FOR I=1 TO N
190 LET S1=S1+ARR(I)^2:LET S2=S2+ARR(I)
200 NEXT
210 LET STDEV=SQR((N*S1-S2^2)/N^2)
220 END DEF
230 FOR J=1 TO N
240 PRINT J;"item =";ARR(J),"standard dev =";STDEV(J)
250 NEXT
260 DATA 2,4,4,4,5,5,7,9</syntaxhighlight>
=={{header|J}}==
J is block-oriented; it expresses algorithms with the semantics of all the data being available at once. It does not have native lexical closure or coroutine semantics. It is possible to implement these semantics in J; see [[Moving Average]] for an example. We will not reprise that here.
<
dev=: - mean
stddevP=: [: %:@mean *:@dev NB. A) 3 equivalent defs for stddevP
Line 1,276 ⟶ 2,173:
stddevP\ 2 4 4 4 5 5 7 9
0 1 0.942809 0.866025 0.979796 1 1.39971 2</
'''Alternatives:'''<br>
Using verbose names for J primitives.
<
sqrt =: %:
sum =: +/
Line 1,290 ⟶ 2,187:
stddevP\ 2 4 4 4 5 5 7 9
0 1 0.942809 0.866025 0.979796 1 1.39971 2</
{{trans|R}}<BR>
Or we could take a cue from the [[#R|R implementation]] and reverse the Bessel correction to stddev:
<
(%:@:(%~<:)@:# * stddev)\ 2 4 4 4 5 5 7 9
0 1 0.942809 0.866025 0.979796 1 1.39971 2</
=={{header|Java}}==
<
int n = 0;
double sum = 0;
Line 1,321 ⟶ 2,218:
}
}
}</
=={{header|JavaScript}}==
===Imperative===
Uses a closure.
<
var n = 0;
var sum = 0.0;
Line 1,344 ⟶ 2,244:
// using WSH
WScript.Echo(stddev.join(', ');</
{{out}}
<pre>0, 1, 0.942809041582063, 0.866025403784439, 0.979795897113273, 1, 1.39970842444753, 2</pre>
===Functional===
====ES5====
Accumulating across a fold
<syntaxhighlight lang="javascript">(function (xs) {
return xs.reduce(function (a, x, i) {
var n = i + 1,
sum_ = a.sum + x,
squaresSum_ = a.squaresSum + (x * x);
return {
sum: sum_,
squaresSum: squaresSum_,
stages: a.stages.concat(
Math.sqrt((squaresSum_ / n) - Math.pow((sum_ / n), 2))
)
};
}, {
sum: 0,
squaresSum: 0,
stages: []
}).stages
})([2, 4, 4, 4, 5, 5, 7, 9]);</syntaxhighlight>
{{Out}}
<syntaxhighlight lang="javascript">[0, 1, 0.9428090415820626, 0.8660254037844386,
0.9797958971132716, 1, 1.3997084244475297, 2]</syntaxhighlight>
====ES6====
As a map-accumulation:
<syntaxhighlight lang="javascript">(() => {
'use strict';
// ---------- CUMULATIVE STANDARD DEVIATION ----------
// cumulativeStdDevns :: [Float] -> [Float]
const cumulativeStdDevns = ns => {
const go = ([s, q]) =>
([i, x]) => {
const
_s = s + x,
_q = q + (x * x),
j = 1 + i;
return [
[_s, _q],
Math.sqrt(
(_q / j) - Math.pow(_s / j, 2)
)
];
};
return mapAccumL(go)([0, 0])(ns)[1];
};
// ---------------------- TEST -----------------------
const main = () =>
showLog(
cumulativeStdDevns([
2, 4, 4, 4, 5, 5, 7, 9
])
);
// --------------------- GENERIC ---------------------
// mapAccumL :: (acc -> x -> (acc, y)) -> acc -> [x] -> (acc, [y])
const mapAccumL = f =>
// A tuple of an accumulation and a list
// obtained by a combined map and fold,
// with accumulation from left to right.
acc => xs => [...xs].reduce((a, x, i) => {
const pair = f(a[0])([i, x]);
return [pair[0], a[1].concat(pair[1])];
}, [acc, []]);
// showLog :: a -> IO ()
const showLog = (...args) =>
console.log(
args
.map(x => JSON.stringify(x, null, 2))
.join(' -> ')
);
// MAIN ---
return main();
})();</syntaxhighlight>
{{Out}}
<pre>[
0,
1,
0.9428090415820626,
0.8660254037844386,
0.9797958971132716,
1,
1.3997084244475297,
2
]</pre>
=={{header|jq}}==
====Observations from a file or array====
We first define a filter, "simulate", that, if given a file of
observations, will emit the standard deviations of the observations
seen so far.
The current state is stored in a JSON object, with this structure:
{ "n": _, "ssd": _, "mean": _ }
Line 1,364 ⟶ 2,366:
where SSD is the sum of squared deviations about the mean.
<
# seen so far, given the current state as input:
def standard_deviation: .ssd / .n | sqrt;
Line 1,389 ⟶ 2,391:
# Begin:
simulate</
'''Example 1'''
# observations.txt
Line 1,401 ⟶ 2,403:
9
{{Out}}
<syntaxhighlight lang="sh">
$ jq -s -f Dynamic_standard_deviation.jq observations.txt
0
Line 1,411 ⟶ 2,413:
1.3997084244475302
1.9999999999999998
</syntaxhighlight>
====Observations from a stream====
Recent versions of jq (after 1.4) support retention of state while processing a stream. This means that any generator (including generators that produce items indefinitely) can be used as the source of observations, without first having to capture all the observations, e.g. in a file or array.
<
def simulate(stream):
foreach stream as $observation
(initial_state;
update_state($observation);
standard_deviation);</
'''Example 2''':
simulate( range(0;10) )
Line 1,438 ⟶ 2,440:
The definitions of the filters update_state/1 and initial_state/0 are as above but are repeated so that this script is self-contained.
<
# jq is assumed to be on PATH
Line 1,475 ⟶ 2,477:
sed -n 1p <<< "$result"
state=$(sed -n 2p <<< "$result")
done</
'''Example 3'''
<
Next observation: 10
0
Line 1,484 ⟶ 2,486:
Next observation: 0
8.16496580927726
</syntaxhighlight>
=={{header|Julia}}==
Use a closure to create a running standard deviation function.
<syntaxhighlight lang="julia">function makerunningstd(::Type{T} = Float64) where T
∑x = ∑x² = zero(T)
n = 0
function runningstd(x)
∑x += x
∑x² += x ^ 2
n += 1
s = ∑x² / n - (∑x / n) ^ 2
return s
end
return runningstd
end
test = Float64[2, 4, 4, 4, 5, 5, 7, 9]
rstd = makerunningstd()
println("Perform a running standard deviation of ", test)
for i in test
println(" - add $i → ", rstd(i))
end</syntaxhighlight>
{{out}}
<pre>Perform a running standard deviation of [2.0, 4.0, 4.0, 4.0, 5.0, 5.0, 7.0, 9.0]
- add 2.0 → 0.0
- add 4.0 → 1.0
- add 4.0 → 0.8888888888888875
- add 4.0 → 0.75
- add 5.0 → 0.9600000000000009
- add 5.0 → 1.0
- add 7.0 → 1.9591836734693864
- add 9.0 → 4.0
</pre>
=={{header|Kotlin}}==
{{trans|Java}}
Using a class to keep the running sum, sum of squares and number of elements added so far:
<syntaxhighlight lang="scala">// version 1.0.5-2
class CumStdDev {
private var n = 0
private var sum = 0.0
private var sum2 = 0.0
fun sd(x: Double): Double {
n++
sum += x
sum2 += x * x
return Math.sqrt(sum2 / n - sum * sum / n / n)
}
}
fun main(args: Array<String>) {
val testData = doubleArrayOf(2.0, 4.0, 4.0, 4.0, 5.0, 5.0, 7.0, 9.0)
val csd = CumStdDev()
for (d in testData) println("Add $d => ${csd.sd(d)}")
}</syntaxhighlight>
{{out}}
<pre>
Add 2.0 => 0.0
Add 4.0 => 1.0
Add 4.0 => 0.9428090415820626
Add 4.0 => 0.8660254037844386
Add 5.0 => 0.9797958971132708
Add 5.0 => 1.0
Add 7.0 => 1.399708424447531
Add 9.0 => 2.0
</pre>
=={{header|Liberty BASIC}}==
Using a global array to maintain the state. Implements definition explicitly.
<syntaxhighlight lang="lb">
dim SD.storage$( 100) ' can call up to 100 versions, using ID to identify.. arrays are global.
' holds (space-separated) number of data items so far, current sum.of.values and current sum.of.squares
Line 1,515 ⟶ 2,588:
Data 2, 4, 4, 4, 5, 5, 7, 9
</syntaxhighlight>
<pre>
New data 2 so S.D. now = 0.000000
Line 1,525 ⟶ 2,598:
New data 7 so S.D. now = 1.399708
New data 9 so S.D. now = 2.000000
</pre>
=={{header|Lobster}}==
<syntaxhighlight lang="lobster">
// Stats computes a running mean and variance
// See Knuth TAOCP vol 2, 3rd edition, page 232
class Stats:
M = 0.0
S = 0.0
n = 0
def incl(x):
n += 1
if n == 1:
M = x
else:
let mm = (x - M)
M += mm / n
S += mm * (x - M)
def mean(): return M
//def variance(): return (if n > 1.0: S / (n - 1.0) else: 0.0) // Bessel's correction
def variance(): return (if n > 0.0: S / n else: 0.0)
def stddev(): return sqrt(variance())
def count(): return n
def test_stdv() -> float:
let v = [2,4,4,4,5,5,7,9]
let s = Stats {}
for(v) x: s.incl(x+0.0)
print concat_string(["Mean: ", string(s.mean()), ", Std.Deviation: ", string(s.stddev())], "")
test_stdv()
</syntaxhighlight>
{{out}}
<pre>
Mean: 5.0, Std.Deviation: 2.0
</pre>
=={{header|Lua}}==
Uses a closure. Translation of JavaScript.
<
local sum, sumsq, k = 0,0,0
return function(n)
Line 1,540 ⟶ 2,649:
for i, v in ipairs{2,4,4,4,5,5,7,9} do
print(ldev(v))
end</
=={{header|Mathematica}}/{{header|Wolfram Language}}==
<syntaxhighlight lang="mathematica">runningSTDDev[n_] := (If[Not[ValueQ[$Data]], $Data = {}];StandardDeviation[AppendTo[$Data, n]])</syntaxhighlight>
=={{header|MATLAB}} / {{header|Octave}}==
The simple form is, computing only the standand deviation of the whole data set:
<
n = length (x);
Line 1,556 ⟶ 2,663:
x2 = mean (x .* x);
dev= sqrt (x2 - m * m)
dev = 2 </
When the intermediate results are also needed, one can use this vectorized form:
<
x2= cumsum(x.^2) ./ [1:n]; % running squares
Line 1,567 ⟶ 2,674:
0.00000 1.00000 0.94281 0.86603 0.97980 1.00000 1.39971 2.00000
</syntaxhighlight>
Here is a vectorized one line solution as a function
<syntaxhighlight lang="matlab">
function stdDevEval(n)
disp(sqrt(sum((n-sum(n)/length(n)).^2)/length(n)));
end
</syntaxhighlight>
=={{header|MiniScript}}==
<syntaxhighlight lang="miniscript">StdDeviator = {}
StdDeviator.count = 0
StdDeviator.sum = 0
StdDeviator.sumOfSquares = 0
StdDeviator.add = function(x)
self.count = self.count + 1
self.sum = self.sum + x
self.sumOfSquares = self.sumOfSquares + x*x
end function
StdDeviator.stddev = function()
m = self.sum / self.count
return sqrt(self.sumOfSquares / self.count - m*m)
end function
sd = new StdDeviator
for x in [2, 4, 4, 4, 5, 5, 7, 9]
sd.add x
end for
print sd.stddev</syntaxhighlight>
{{out}}
<pre>2</pre>
=={{header|МК-61/52}}==
<syntaxhighlight lang="text">0 П4 П5 П6 С/П П0 ИП5 + П5 ИП0
x^2 ИП6 + П6 КИП4 ИП6 ИП4 / ИП5 ИП4
/ x^2 - КвКор БП 04</
Instruction: В/О С/П ''number'' С/П ''number'' С/П ...
=={{header|
{{trans|Java}}
<syntaxhighlight lang="nanoquery">class StdDev
declare n
declare sum
declare sum2
def StdDev()
n = 0
sum = 0
sum2 = 0
end
def sd(x)
this.n += 1
this.sum += x
this.sum2 += x*x
return sqrt(sum2/n - sum*sum/n/n)
end
end
testData = {2,4,4,4,5,5,7,9}
sd = new(StdDev)
for x in testData
println sd.sd(x)
end</syntaxhighlight>
{{out}}
<pre>0.0
1.0
0.9428090415820634
0.8660254037844386
0.9797958971132712
1.0
1.3997084244475304
2.0</pre>
=={{header|Nim}}==
===Using global variables===
<syntaxhighlight lang="nim">import math, strutils
var sdSum, sdSum2, sdN = 0.0
proc sd(x: float): float =
sdN += 1
sdSum += x
sdSum2 += x * x
sqrt(sdSum2 / sdN - sdSum * sdSum / (sdN * sdN))
for value in [float 2,4,4,4,5,5,7,9]:
echo value, " ", formatFloat(sd(value), precision = -1)</syntaxhighlight>
{{out}}
<pre>2 0
4 1
4 0.942809
Line 1,598 ⟶ 2,779:
7 1.39971
9 2</pre>
===Using an accumulator object===
<syntaxhighlight lang="nim">import math, strutils
type SDAccum = object
sdN, sdSum, sdSum2: float
var accum: SDAccum
proc add(accum: var SDAccum; value: float): float =
# Add a value to the accumulator. Return the standard deviation.
accum.sdN += 1
accum.sdSum += value
accum.sdSum2 += value * value
result = sqrt(accum.sdSum2 / accum.sdN - accum.sdSum * accum.sdSum / (accum.sdN * accum.sdN))
for value in [float 2, 4, 4, 4, 5, 5, 7, 9]:
echo value, " ", formatFloat(accum.add(value), precision = -1)</syntaxhighlight>
{{out}}
Same output.
===Using a closure===
<syntaxhighlight lang="nim">import math, strutils
func accumBuilder(): auto =
var sdSum, sdSum2, sdN = 0.0
result = func(value: float): float =
sdN += 1
sdSum += value
sdSum2 += value * value
result = sqrt(sdSum2 / sdN - sdSum * sdSum / (sdN * sdN))
let std = accumBuilder()
for value in [float 2, 4, 4, 4, 5, 5, 7, 9]:
echo value, " ", formatFloat(std(value), precision = -1)</syntaxhighlight>
{{out}}
Same output.
=={{header|Objeck}}==
{{trans|Java}}
<syntaxhighlight lang="objeck">
use Structure;
bundle Default {
class StdDev {
nums : FloatVector;
New() {
nums := FloatVector->New();
}
function : Main(args : String[]) ~ Nil {
sd := StdDev->New();
test_data := [2.0, 4.0, 4.0, 4.0, 5.0, 5.0, 7.0, 9.0];
each(i : test_data) {
sd->AddNum(test_data[i]);
sd->GetSD()->PrintLine();
};
}
method : public : AddNum(num : Float) ~ Nil {
nums->AddBack(num);
}
method : public : native : GetSD() ~ Float {
sq_diffs := 0.0;
avg := nums->Average();
each(i : nums) {
num := nums->Get(i);
sq_diffs += (num - avg) * (num - avg);
};
return (sq_diffs / nums->Size())->SquareRoot();
}
}
}
</syntaxhighlight>
=={{header|Objective-C}}==
<
@interface SDAccum : NSObject
Line 1,653 ⟶ 2,916:
}
return 0;
}</
===Blocks===
Line 1,659 ⟶ 2,922:
{{works with|iOS|4+}}
<
typedef double (^Func)(double); // a block that takes a double and returns a double
Line 1,688 ⟶ 2,951:
}
return 0;
}</
=={{header|OCaml}}==
<
let stddev l =
Line 1,745 ⟶ 2,968:
Printf.printf "List: ";
List.iter (Printf.printf "%g ") l;
Printf.printf "\nStandard deviation: %g\n" (stddev l)</
{{out}}
<pre>
List: 2 4 4 4 5 5 7 9
Standard deviation: 2
</pre>
=={{header|Oforth}}==
Oforth does not have global variables that can be used to create statefull functions.
Here, we create a channel to hold current list of numbers. Constraint is that this channel can't hold mutable objects. On the other hand, stddev function is thread safe and can be called by tasks running in parallel.
<syntaxhighlight lang="oforth">Channel new [ ] over send drop const: StdValues
: stddev(x)
| l |
StdValues receive x + dup ->l StdValues send drop
#qs l map sum l size asFloat / l avg sq - sqrt ;</syntaxhighlight>
{{out}}
<pre>
>[ 2, 4, 4, 4, 5, 5, 7, 9 ] apply(#[ stddev println ])
0
1
0.942809041582063
0.866025403784439
0.979795897113272
1
1.39970842444753
2
ok
>
</pre>
=={{header|ooRexx}}==
{{works with|oorexx}}
<
x = .array~of(2,4,4,4,5,5,7,9)
sd = 0
do i = 1 to x~size
sd = sdacc~value(x[i])
Say '#'i 'value =' x[i] 'stdev =' sd
end
::class SDAccum
Line 1,798 ⟶ 3,047:
ans = ( prev + ( n / prev ) ) / 2
end
return ans</
{{out}}
<pre>#1 value = 2 stdev = 0
#2 value = 4 stdev = 1
#3 value = 4 stdev = 0.94280905
#4 value = 4 stdev = 0.866025405
#5 value = 5 stdev = 0.979795895
#6 value = 5 stdev = 1
#7 value = 7 stdev = 1.39970844
#8 value = 9 stdev = 2</pre>
=={{header|PARI/GP}}==
Uses the Cramer-Young updating algorithm. For demonstration it displays the mean and variance at each step.
<
myT=x;
myS=0;
Line 1,820 ⟶ 3,078:
print("Standard deviation: ",sqrt(myS/myN))
};
addpoints([2,4,4,4,5,5,7,9])</
=={{header|Pascal}}==
===Std.Pascal===
{{trans|AWK}}
<
uses math;
const
Line 1,851 ⟶ 3,110:
writeln(i,' item=',arr[i]:2:0,' stddev=',stddev(i):18:15)
end
end.</
{{out}}
<pre>1 item= 2 stddev= 0.000000000000000
Line 1,861 ⟶ 3,120:
7 item= 7 stddev= 1.399708424447530
8 item= 9 stddev= 2.000000000000000</pre>
==={{header|Delphi}}===
<syntaxhighlight lang="delphi">program prj_CalcStdDerv;
{$APPTYPE CONSOLE}
uses
Math;
var Series:Array of Extended;
UserString:String;
function AppendAndCalc(NewVal:Extended):Extended;
begin
setlength(Series,high(Series)+2);
Series[high(Series)] := NewVal;
result := PopnStdDev(Series);
end;
const data:array[0..7] of Extended =
(2,4,4,4,5,5,7,9);
var rr: Extended;
begin
setlength(Series,0);
for rr in data do
begin
writeln(rr,' -> ',AppendAndCalc(rr));
end;
Readln;
end. </syntaxhighlight>
{{out}}
<pre>
2.0000000000000000E+0000 -> 0.0000000000000000E+0000
4.0000000000000000E+0000 -> 1.0000000000000000E+0000
4.0000000000000000E+0000 -> 9.4280904158206337E-0001
4.0000000000000000E+0000 -> 8.6602540378443865E-0001
5.0000000000000000E+0000 -> 9.7979589711327124E-0001
5.0000000000000000E+0000 -> 1.0000000000000000E+0000
7.0000000000000000E+0000 -> 1.3997084244475303E+0000
9.0000000000000000E+0000 -> 2.0000000000000000E+0000
</pre>
=={{header|Perl}}==
<
package SDAccum;
sub new {
Line 1,899 ⟶ 3,201:
return $self->stddev;
}
}</
<
my $sd;
Line 1,907 ⟶ 3,209:
$sd = $sdacc->value($v);
}
print "std dev = $sd\n";</
A much shorter version using a closure and a property of the variance:
<
{
my $num, $sum, $sum2;
Line 1,924 ⟶ 3,226:
}
print stddev($_), "\n" for qw(2 4 4 4 5 5 7 9);</
{{out}}
Line 1,936 ⟶ 3,238:
2</pre>
one-liner:
<syntaxhighlight lang="bash">perl -MMath::StdDev -e '$d=new Math::StdDev;foreach my $v ( 2,4,4,4,5,5,7,9 ) {$d->Update($v); print $d->variance(),"\n"}'</syntaxhighlight>
small script:
<syntaxhighlight lang="perl">use Math::StdDev;
$d=new Math::StdDev;
foreach my $v ( 2,4,4,4,5,5,7,9 ) {
$d->Update($v);
print $d->variance(),"\n"
}</syntaxhighlight>
{{out}}
<pre>
0
1
0.942809041582063
Line 1,973 ⟶ 3,260:
2</pre>
=={{header|
demo\rosetta\Standard_deviation.exw contains a copy of this code and a version that could be the basis for a library version that can handle multiple active data sets concurrently.
<!--<syntaxhighlight lang="phix">(phixonline)-->
<span style="color: #008080;">with</span> <span style="color: #008080;">javascript_semantics</span>
<span style="color: #004080;">atom</span> <span style="color: #000000;">sdn</span> <span style="color: #0000FF;">=</span> <span style="color: #000000;">0</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">sdsum</span> <span style="color: #0000FF;">=</span> <span style="color: #000000;">0</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">sdsumsq</span> <span style="color: #0000FF;">=</span> <span style="color: #000000;">0</span>
<span style="color: #008080;">procedure</span> <span style="color: #000000;">sdadd</span><span style="color: #0000FF;">(</span><span style="color: #004080;">atom</span> <span style="color: #000000;">n</span><span style="color: #0000FF;">)</span>
<span style="color: #000000;">sdn</span> <span style="color: #0000FF;">+=</span> <span style="color: #000000;">1</span>
<span style="color: #000000;">sdsum</span> <span style="color: #0000FF;">+=</span> <span style="color: #000000;">n</span>
<span style="color: #000000;">sdsumsq</span> <span style="color: #0000FF;">+=</span> <span style="color: #000000;">n</span><span style="color: #0000FF;">*</span><span style="color: #000000;">n</span>
<span style="color: #008080;">end</span> <span style="color: #008080;">procedure</span>
<span style="color: #008080;">function</span> <span style="color: #000000;">sdavg</span><span style="color: #0000FF;">()</span>
<span style="color: #008080;">return</span> <span style="color: #000000;">sdsum</span><span style="color: #0000FF;">/</span><span style="color: #000000;">sdn</span>
<span style="color: #008080;">end</span> <span style="color: #008080;">function</span>
<span style="color: #008080;">function</span> <span style="color: #000000;">sddev</span><span style="color: #0000FF;">()</span>
<span style="color: #008080;">return</span> <span style="color: #7060A8;">sqrt</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sdsumsq</span><span style="color: #0000FF;">/</span><span style="color: #000000;">sdn</span> <span style="color: #0000FF;">-</span> <span style="color: #7060A8;">power</span><span style="color: #0000FF;">(</span><span style="color: #000000;">sdsum</span><span style="color: #0000FF;">/</span><span style="color: #000000;">sdn</span><span style="color: #0000FF;">,</span><span style="color: #000000;">2</span><span style="color: #0000FF;">))</span>
<span style="color: #008080;">end</span> <span style="color: #008080;">function</span>
<span style="color: #000080;font-style:italic;">--test code:</span>
<span style="color: #008080;">constant</span> <span style="color: #000000;">testset</span> <span style="color: #0000FF;">=</span> <span style="color: #0000FF;">{</span><span style="color: #000000;">2</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">4</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">4</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">4</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">5</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">5</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">7</span><span style="color: #0000FF;">,</span> <span style="color: #000000;">9</span><span style="color: #0000FF;">}</span>
<span style="color: #004080;">integer</span> <span style="color: #000000;">ti</span>
<span style="color: #008080;">for</span> <span style="color: #000000;">i</span><span style="color: #0000FF;">=</span><span style="color: #000000;">1</span> <span style="color: #008080;">to</span> <span style="color: #7060A8;">length</span><span style="color: #0000FF;">(</span><span style="color: #000000;">testset</span><span style="color: #0000FF;">)</span> <span style="color: #008080;">do</span>
<span style="color: #000000;">ti</span> <span style="color: #0000FF;">=</span> <span style="color: #000000;">testset</span><span style="color: #0000FF;">[</span><span style="color: #000000;">i</span><span style="color: #0000FF;">]</span>
<span style="color: #000000;">sdadd</span><span style="color: #0000FF;">(</span><span style="color: #000000;">ti</span><span style="color: #0000FF;">)</span>
<span style="color: #7060A8;">printf</span><span style="color: #0000FF;">(</span><span style="color: #000000;">1</span><span style="color: #0000FF;">,</span><span style="color: #008000;">"N=%d Item=%d Avg=%5.3f StdDev=%5.3f\n"</span><span style="color: #0000FF;">,{</span><span style="color: #000000;">i</span><span style="color: #0000FF;">,</span><span style="color: #000000;">ti</span><span style="color: #0000FF;">,</span><span style="color: #000000;">sdavg</span><span style="color: #0000FF;">(),</span><span style="color: #000000;">sddev</span><span style="color: #0000FF;">()})</span>
<span style="color: #008080;">end</span> <span style="color: #008080;">for</span>
<!--</syntaxhighlight>-->
{{out}}
<pre>
N=1 Item=2 Avg=2.000 StdDev=0.000
N=2 Item=4 Avg=3.000 StdDev=1.000
N=3 Item=4 Avg=3.333 StdDev=0.943
N=4 Item=4 Avg=3.500 StdDev=0.866
N=5 Item=5 Avg=3.800 StdDev=0.980
N=6 Item=5 Avg=4.000 StdDev=1.000
N=7 Item=7 Avg=4.429 StdDev=1.400
N=8 Item=9 Avg=5.000 StdDev=2.000
</pre>
=={{header|PHP}}==
This is just straight PHP class usage, respecting the specifications "stateful" and "one at a time":
<syntaxhighlight lang="php"><?php
class sdcalc {
private $cnt, $sumup, $square;
function __construct() {
$this->reset();
}
# callable on an instance
function reset() {
$this->cnt=0; $this->sumup=0; $this->square=0;
}
function add($f) {
$this->cnt++;
$this->sumup += $f;
$this->square += pow($f, 2);
return $this->calc();
}
function calc() {
if ($this->cnt==0 || $this->sumup==0) {
return 0;
} else {
return sqrt($this->square / $this->cnt - pow(($this->sumup / $this->cnt),2));
}
}
}
# start test, adding test data one by one
$c = new sdcalc();
foreach ([2,4,4,4,5,5,7,9] as $v) {
printf('Adding %g: result %g%s', $v, $c->add($v), PHP_EOL);
}</syntaxhighlight>
This will produce the output:
<pre>Adding 2: result 0
Adding 4: result 1
Adding 4: result 0.942809
Adding 4: result 0.866025
Adding 5: result 0.979796
Adding 5: result 1
Adding 7: result 1.39971
Adding 9: result 2
</pre>
=={{header|PicoLisp}}==
<
(de stdDev ()
Line 2,021 ⟶ 3,366:
(let Fun (stdDev)
(for N (2.0 4.0 4.0 4.0 5.0 5.0 7.0 9.0)
(prinl (format N *Scl) " -> " (format (Fun N) *Scl)) ) )</
{{out}}
<pre>2.00 -> 0.00
4.00 -> 1.00
Line 2,031 ⟶ 3,376:
7.00 -> 1.40
9.00 -> 2.00</pre>
=={{header|PL/I}}==
<syntaxhighlight lang="pli">*process source attributes xref;
stddev: proc options(main);
declare a(10) float init(1,2,3,4,5,6,7,8,9,10);
declare stdev float;
declare i fixed binary;
stdev=std_dev(a);
put skip list('Standard deviation', stdev);
std_dev: procedure(a) returns(float);
declare a(*) float, n fixed binary;
n=hbound(a,1);
begin;
declare b(n) float, average float;
declare i fixed binary;
do i=1 to n;
b(i)=a(i);
end;
average=sum(a)/n;
put skip data(average);
return( sqrt(sum(b**2)/n - average**2) );
end;
end std_dev;
end;</syntaxhighlight>
{{out}}
<pre>AVERAGE= 5.50000E+0000;
Standard deviation 2.87228E+0000 </pre>
=={{header|PowerShell}}==
This implementation takes the form of an advanced function
which can act like a cmdlet and receive input from the pipeline.
<syntaxhighlight lang="powershell">function Get-StandardDeviation {
begin {
$avg = 0
Line 2,046 ⟶ 3,422:
[Math]::Sqrt($sum / $nums.Length)
}
}</
Usage as follows:
<pre>PS> 2,4,4,4,5,5,7,9 | Get-StandardDeviation
Line 2,059 ⟶ 3,435:
=={{header|PureBasic}}==
<
Declare.d Standard_deviation(x)
Line 2,087 ⟶ 3,463:
MyList:
Data.i 2,4,4,4,5,5,7,9
EndDataSection</
{{out}}
<pre>
0.0000000000
1.0000000000
Line 2,098 ⟶ 3,475:
1.3997084244
2.0000000000
</pre>
=={{header|Python}}==
===Python: Using a function with attached properties===
The program should work with Python 2.x and 3.x,
although the output would not be a tuple in 3.x
<syntaxhighlight lang="python">>>> from math import sqrt
>>> def sd(x):
sd.sum += x
Line 2,123 ⟶ 3,502:
(7, 1.3997084244475311)
(9, 2.0)
>>></
===Python: Using a class instance===
<
def __init__(self):
self.sum, self.sum2, self.n = (0,0,0)
Line 2,138 ⟶ 3,517:
>>> sd_inst = SD()
>>> for value in (2,4,4,4,5,5,7,9):
print (value, sd_inst.sd(value))</
===
You could rename the method <code>sd</code> to <code>__call__</code> this would make the class instance callable like a function so instead of using <code>sd_inst.sd(value)</code> it would change to <code>sd_inst(value)</code> for the same results.
===Python: Using a Closure===
{{Works with|Python|3.x}}
<
>>> def sdcreator():
sum = sum2 = n = 0
Line 2,166 ⟶ 3,548:
5 1.0
7 1.39970842445
9 2.0</
===Python: Using an extended generator===
{{Works with|Python|2.5+}}
<
>>> def sdcreator():
sum = sum2 = n = 0
Line 2,193 ⟶ 3,575:
5 1.0
7 1.39970842445
9 2.0</
===Python: In a couple of 'functional' lines===
<syntaxhighlight lang="python">>>> myMean = lambda MyList : reduce(lambda x, y: x + y, MyList) / float(len(MyList))
>>> myStd = lambda MyList : (reduce(lambda x,y : x + y , map(lambda x: (x-myMean(MyList))**2 , MyList)) / float(len(MyList)))**.5
>>> print myStd([2,4,4,4,5,5,7,9])
2.0
</syntaxhighlight>
=={{header|R}}==
To compute the running sum, one must keep track of the number of items, the sum of values, and the sum of squares.
If the goal is to get a vector of running standard deviations, the simplest is to do it with cumsum:
<syntaxhighlight lang="rsplus">cumsd <- function(x) {
sqrt(cumsum(x^2) / n - (cumsum(x) / n)^2)
}
set.seed(12345L)
x <- rnorm(10)
cumsd(x)
# [1] 0.0000000 0.3380816 0.8752973 1.1783628 1.2345538 1.3757142 1.2867220 1.2229056 1.1665168 1.1096814
# Compare to the naive implementation, i.e. compute sd on each sublist:
Vectorize(function(k) sd(x[1:k]) * sqrt((k - 1) / k))(seq_along(x))
# [1] NA 0.3380816 0.8752973 1.1783628 1.2345538 1.3757142 1.2867220 1.2229056 1.1665168 1.1096814
# Note that the first is NA because sd is unbiased formula, hence there is a division by n-1, which is 0 for n=1.</syntaxhighlight>
The task requires an accumulator solution:
<syntaxhighlight lang="rsplus">accumsd <- function() {
n <- 0
m <- 0
s <- 0
function(x) {
n <<- n + 1
m <<- m + x
s <<- s + x * x
sqrt(s / n - (m / n)^2)
}
}
f <- accumsd()
sapply(x, f)
# [1] 0.0000000 0.3380816 0.8752973 1.1783628 1.2345538 1.3757142 1.2867220 1.2229056 1.1665168 1.1096814</syntaxhighlight>
=={{header|Racket}}==
<
#lang racket
(require math)
Line 2,231 ⟶ 3,636:
;; run it on each number, return the last result
(last (map running-stddev '(2 4 4 4 5 5 7 9)))
</syntaxhighlight>
=={{header|
(formerly Perl 6)
Using a closure:
<syntaxhighlight lang="raku" line>sub sd (@a) {
my $mean = @a R/ [+] @a;
sqrt @a R/ [+] map (* - $mean)², @a;
}
sub sdaccum {
my @a;
return { push @a, $^x; sd @a; };
}
my &f = sdaccum;
say f $_ for 2, 4, 4, 4, 5, 5, 7, 9;</syntaxhighlight>
Using a state variable (remember that <tt><(x-<x>)²> = <x²> - <x>²</tt>):
<syntaxhighlight lang="raku" line>sub stddev($x) {
sqrt
( .[2] += $x²) / ++.[0]
- ((.[1] += $x ) / .[0])²
given state @;
}
say .&stddev for <2 4 4 4 5 5 7 9>;</syntaxhighlight>
{{out}}
<pre>0
1
0.942809041582063
0.866025403784439
0.979795897113271
1
1.39970842444753
2</pre>
=={{header|REXX}}==
These REXX versions use ''running sums''.
===show running sums===
<syntaxhighlight lang="rexx">/*REXX program calculates and displays the standard deviation of a given set of numbers.*/
parse arg # /*obtain optional arguments from the CL*/
if #='' then #= 2 4 4 4 5 5 7 9 /*None specified? Then use the default*/
n= words(#); $= 0; $$= 0; L= length(n) /*N: # items; $,$$: sums to be zeroed*/
/* [↓] process each number in the list*/
do j=1 for n
_= word(#, j); $= $ + _
$$= $$ + _**2
say ' item' right(j, L)":" right(_, 4) ' average=' left($/j, 12),
' standard deviation=' sqrt($$/j - ($/j)**2)
end /*j*/ /* [↑] prettify output with whitespace*/
say 'standard deviation: ' sqrt($$/n - ($/n)**2) /*calculate & display the std deviation*/
exit 0 /*stick a fork in it, we're all done. */
/*──────────────────────────────────────────────────────────────────────────────────────*/
sqrt: procedure; parse arg x; if x=0 then return 0; d=digits(); h=d+6; m.=9; numeric form
numeric digits; parse value format(x,2,1,,0) 'E0' with g 'E' _ .; g=g * .5'e'_ % 2
do j=0 while h>9; m.j=h; h=h%2+1; end /*j*/
do k=j+5 to 0 by -1; numeric digits m.k; g=(g+x/g)*.5; end /*k*/
numeric digits d; return g/1</syntaxhighlight>
{{out|output|text= when using the default input of: <tt> 2 4 4 4 5 5 7 9 </tt>}}
<pre>
item 1: 2 average= 2 standard deviation= 0
Line 2,262 ⟶ 3,707:
item 7: 7 average= 4.42857143 standard deviation= 1.39970843
item 8: 9 average= 5 standard deviation= 2
standard deviation: 2
</pre>
===only show standard deviation===
<syntaxhighlight lang="rexx">/*REXX program calculates and displays the standard deviation of a given set of numbers.*/
parse arg # /*obtain optional arguments from the CL*/
if #='' then #= 2 4 4 4 5 5 7 9 /*None specified? Then use the default*/
n= words(#); $= 0; $$= 0 /*N: # items; $,$$: sums to be zeroed*/
/* [↓] process each number in the list*/
do j=1 for n /*perform summation on two sets of #'s.*/
_= word(#, j); $= $ + _
$$= $$ + _**2
end /*j*/
say 'standard deviation: ' sqrt($$/n - ($/n)**2) /*calculate&display the std, deviation.*/
exit 0 /*stick a fork in it, we're all done. */
/*──────────────────────────────────────────────────────────────────────────────────────*/
sqrt: procedure; parse arg x; if x=0 then return 0; d=digits(); h=d+6; m.=9; numeric form
numeric digits; parse value format(x,2,1,,0) 'E0' with g 'E' _ .; g=g * .5'e'_ % 2
do j=0 while h>9; m.j=h; h=h%2+1; end /*j*/
do k=j+5 to 0 by -1; numeric digits m.k; g=(g+x/g)*.5; end /*k*/
numeric digits d; return g/1</syntaxhighlight>
{{out|output|text= when using the default input of: <tt> 2 4 4 4 5 5 7 9 </tt>}}
<pre>
standard deviation: 2
</pre>
=={{header|Ring}}==
<syntaxhighlight lang="ring">
# Project : Cumulative standard deviation
decimals(6)
sdsave = list(100)
sd = "2,4,4,4,5,5,7,9"
sumval = 0
sumsqs = 0
for num = 1 to 8
sd = substr(sd, ",", "")
stddata = number(sd[num])
sumval = sumval + stddata
sumsqs = sumsqs + pow(stddata,2)
standdev = pow(((sumsqs / num) - pow((sumval /num),2)),0.5)
sdsave[num] = string(num) + " " + string(sumval) +" " + string(sumsqs)
see "" + num + " value in = " + stddata + " Stand Dev = " + standdev + nl
next
</syntaxhighlight>
Output:
<pre>
1 value in = 2 Stand Dev = 0
2 value in = 4 Stand Dev = 1
3 value in = 4 Stand Dev = 0.942809
4 value in = 4 Stand Dev = 0.866025
5 value in = 5 Stand Dev = 0.979796
6 value in = 5 Stand Dev = 1
7 value in = 7 Stand Dev = 1.399708
8 value in = 9 Stand Dev = 2
</pre>
=={{header|RPL}}==
===Basic RPL===
≪ CL∑ { } SWAP
1 OVER SIZE '''FOR''' j
DUP j GET ∑+
'''IF''' j 1 > '''THEN'''
SDEV ∑DAT SIZE 1 GET DUP 1 - SWAP / √ *
ROT SWAP + SWAP '''END'''
'''NEXT'''
DROP CL∑
≫ '<span style="color:blue>CSDEV</span>' STO
===RPL 1993===
≪ CL∑
1 ≪ ∑+ PSDEV ≫ DOSUBS CL∑
≫ '<span style="color:blue>CSDEV</span>' STO
{{out}}
<pre>
1: { 0 1 0.942809041582 0.866025403784 0.979795897113 1 1.39970842445 2 }
</pre>
Line 2,270 ⟶ 3,791:
"Simplification of the formula [...] for standard deviation [...] can be memorized as taking the square root of (the average of the squares less the square of the average)." [[wp:Standard_deviation#Simplification_of_the_formula|c.f. wikipedia]].
<
def initialize
@n, @sum, @sumofsquares = 0, 0.0, 0.0
Line 2,294 ⟶ 3,815:
sd = StdDevAccumulator.new
i = 0
[2,4,4,4,5,5,7,9].each {|n| puts "adding #{n}: stddev of #{i+=1} samples is #{sd << n}" }</
<pre>adding 2: stddev of 1 samples is 0.0
Line 2,306 ⟶ 3,827:
=== Closure ===
<
n, sum, sum2 = 0, 0.0, 0.0
lambda do |num|
Line 2,317 ⟶ 3,838:
sd = sdaccum
[2,4,4,4,5,5,7,9].each {|n| print sd.call(n), ", "}</
<pre>0.0, 1.0, 0.942809041582063, 0.866025403784439, 0.979795897113272, 1.0, 1.39970842444753, 2.0, </pre>
=={{header|Run BASIC}}==
<
'holds (space-separated) number of data , sum of values and sum of squares
sd$ = "2,4,4,4,5,5,7,9"
Line 2,338 ⟶ 3,858:
print num;" value in = ";stdData; " Stand Dev = "; using("###.######", standDev)
next num</
<pre>1 value in = 2 Stand Dev = 0.000000
2 value in = 4 Stand Dev = 1.000000
Line 2,348 ⟶ 3,868:
8 value in = 9 Stand Dev = 2.000000</pre>
=={{header|Rust}}==
Using a struct:
{{trans|Java}}
<syntaxhighlight lang="rust">pub struct CumulativeStandardDeviation {
n: f64,
sum: f64,
sum_sq: f64
}
impl CumulativeStandardDeviation {
pub fn new() -> Self {
CumulativeStandardDeviation {
n: 0.,
sum: 0.,
sum_sq: 0.
}
}
fn push(&mut self, x: f64) -> f64 {
self.n += 1.;
self.sum += x;
self.sum_sq += x * x;
(self.sum_sq / self.n - self.sum * self.sum / self.n / self.n).sqrt()
}
fn main() {
let mut cum_stdev = CumulativeStandardDeviation::new();
for num in nums.iter() {
println!("{}", cum_stdev.push(*num as f64));
}
}</syntaxhighlight>
{{out}}
<pre>
0
1
0.9428090415820626
0.8660254037844386
0.9797958971132708
1
1.399708424447531
2
</pre>
Using a closure:
<syntaxhighlight lang="rust">fn sd_creator() -> impl FnMut(f64) -> f64 {
let mut n = 0.0;
let mut sum = 0.0;
let mut sum_sq = 0.0;
move |x| {
sum += x;
sum_sq += x*x;
n += 1.0;
(sum_sq / n - sum * sum / n / n).sqrt()
}
}
fn main() {
let nums = [2, 4, 4, 4, 5, 5, 7, 9];
let mut sd_acc = sd_creator();
for num in nums.iter() {
println!("{}", sd_acc(*num as f64));
}
}</syntaxhighlight>
{{out}}
<pre>
0
1
0.9428090415820626
0.8660254037844386
0.9797958971132708
1
1.399708424447531
2
</pre>
=={{header|SAS}}==
<syntaxhighlight lang="sas">
*--Load the test data;
data test1;
Line 2,411 ⟶ 3,982:
var n sd /*mean*/;
run;
</syntaxhighlight>
{{out}}
<pre>
N SD
Line 2,427 ⟶ 3,997:
8 2.00000
</pre>
=={{header|Scala}}==
===Generic for any numeric type===
{{libheader|Scala}}
<syntaxhighlight lang="scala">import scala.math.sqrt
object StddevCalc extends App {
def calcAvgAndStddev[T](ts: Iterable[T])(implicit num: Fractional[T]): (T, Double) = {
def avg(ts: Iterable[T])(implicit num: Fractional[T]): T =
num.div(ts.sum, num.fromInt(ts.size)) // Leaving with type of function T
val mean: T = avg(ts) // Leave val type of T
// Root of mean diffs
val stdDev = sqrt(ts.map { x =>
val diff = num.toDouble(num.minus(x, mean))
diff * diff
}.sum / ts.size)
(mean, stdDev)
}
println(calcAvgAndStddev(List(2.0E0, 4.0, 4, 4, 5, 5, 7, 9)))
println(calcAvgAndStddev(Set(1.0, 2, 3, 4)))
println(calcAvgAndStddev(0.1 to 1.1 by 0.05))
println(calcAvgAndStddev(List(BigDecimal(120), BigDecimal(1200))))
println(s"Successfully completed without errors. [total ${scala.compat.Platform.currentTime - executionStart}ms]")
}</syntaxhighlight>
=={{header|Scheme}}==
<
(define (
(
(lambda (x)
(set! nums (cons x nums))
(let* ((mean (/ (apply + nums) (length nums)))
(mean-sqr (lambda (y) (expt (- y mean) 2)))
(variance (/ (apply + (map mean-sqr nums)) (length nums))))
(sqrt variance)))))
(let loop ((f (standart-deviation-generator))
(input '(2 4 4 4 5 5 7 9)))
(unless (null? input)
(display (f (car input)))
(newline)
(loop f (cdr input))))
</syntaxhighlight>
=={{header|Scilab}}==
Scilab has the built-in function '''stdev''' to compute the standard deviation of a sample so it is straightforward to have the standard deviation of a sample with a correction of the bias.
<syntaxhighlight lang="text">T=[2,4,4,4,5,5,7,9];
stdev(T)*sqrt((length(T)-1)/length(T))</syntaxhighlight>
{{out}}
<pre>-->T=[2,4,4,4,5,5,7,9];
-->stdev(T)*sqrt((length(T)-1)/length(T))
ans = 2.</pre>
=={{header|Sidef}}==
Using an object to keep state:
<syntaxhighlight lang="ruby">class StdDevAccumulator(n=0, sum=0, sumofsquares=0) {
method <<(num) {
n += 1
sum += num
sumofsquares += num**2
self
}
method stddev {
sqrt(sumofsquares/n - pow(sum/n, 2))
}
method to_s {
self.stddev.to_s
}
}
var i = 0
var sd = StdDevAccumulator()
[2,4,4,4,5,5,7,9].each {|n|
say "adding #{n}: stddev of #{i+=1} samples is #{sd << n}"
}</syntaxhighlight>
{{out}}
<pre>
adding 2: stddev of 1 samples is 0
adding 4: stddev of 2 samples is 1
adding 4: stddev of 3 samples is 0.942809041582063365867792482806465385713114583585
adding 4: stddev of 4 samples is 0.866025403784438646763723170752936183471402626905
adding 5: stddev of 5 samples is 0.979795897113271239278913629882356556786378992263
adding 5: stddev of 6 samples is 1
adding 7: stddev of 7 samples is 1.39970842444753034182701947126050936683768427466
adding 9: stddev of 8 samples is 2
</pre>
Using ''static'' variables:
<syntaxhighlight lang="ruby">func stddev(x) {
static(num=0, sum=0, sum2=0)
num++
sqrt(
(sum2 += x**2) / num -
(((sum += x) / num)**2)
)
}
%n(2 4 4 4 5 5 7 9).each { say stddev(_) }</syntaxhighlight>
{{out}}
<pre>
0
1
0.942809041582063365867792482806465385713114583585
0.866025403784438646763723170752936183471402626905
0.979795897113271239278913629882356556786378992263
1
1.39970842444753034182701947126050936683768427466
2
</pre>
=={{header|Smalltalk}}==
{{works with|GNU Smalltalk}}
<
|sum sum2 num|
SDAccum class >> new [ |o|
Line 2,457 ⟶ 4,137:
]
stddev [ ^ (self variance) sqrt ]
].</
<
sdacc := SDAccum new.
#( 2 4 4 4 5 5 7 9 ) do: [ :v | sd := sdacc value: v ].
('std dev = %1' % { sd }) displayNl.</
=={{header|SQL}}==
{{works with|Postgresql}}
<syntaxhighlight lang="sql">-- the minimal table
create table if not exists teststd (n double precision not null);
-- code modularity with view, we could have used a common table expression instead
create view vteststd as
select count(n) as cnt,
sum(n) as tsum,
sum(power(n,2)) as tsqr
from teststd;
-- you can of course put this code into every query
create or replace function std_dev() returns double precision as $$
select sqrt(tsqr/cnt - (tsum/cnt)^2) from vteststd;
$$ language sql;
-- test data is: 2,4,4,4,5,5,7,9
insert into teststd values (2);
select std_dev() as std_deviation;
insert into teststd values (4);
select std_dev() as std_deviation;
insert into teststd values (4);
select std_dev() as std_deviation;
insert into teststd values (4);
select std_dev() as std_deviation;
insert into teststd values (5);
select std_dev() as std_deviation;
insert into teststd values (5);
select std_dev() as std_deviation;
insert into teststd values (7);
select std_dev() as std_deviation;
insert into teststd values (9);
select std_dev() as std_deviation;
-- cleanup test data
delete from teststd;
</syntaxhighlight>
With a command like '''psql <rosetta-std-dev.sql''' you will get an output like this: (duplicate lines generously deleted, locale is DE)
<pre>
CREATE TABLE
FEHLER: Relation »vteststd« existiert bereits
CREATE FUNCTION
INSERT 0 1
std_deviation
---------------
0
(1 Zeile)
INSERT 0 1
std_deviation
---------------
1
0.942809041582063
0.866025403784439
0.979795897113272
1
1.39970842444753
2
DELETE 8
</pre>
=={{header|Swift}}==
<
class stdDev{
Line 2,493 ⟶ 4,234:
}
var aa = stdDev()</
{{out}}
<pre>
value 2 SD = 0.0
Line 2,505 ⟶ 4,246:
value 9 SD = 2.0
</pre>
Functional:
<syntaxhighlight lang="swift">
func standardDeviation(arr : [Double]) -> Double
{
let length = Double(arr.count)
let avg = arr.reduce(0, { $0 + $1 }) / length
let sumOfSquaredAvgDiff = arr.map { pow($0 - avg, 2.0)}.reduce(0, {$0 + $1})
return sqrt(sumOfSquaredAvgDiff / length)
}
let responseTimes = [ 18.0, 21.0, 41.0, 42.0, 48.0, 50.0, 55.0, 90.0 ]
standardDeviation(responseTimes) // 20.8742514835862
standardDeviation([2,4,4,4,5,5,7,9]) // 2.0
</syntaxhighlight>
=={{header|Tcl}}==
===With a Class===
{{works with|Tcl|8.6}} or {{libheader|TclOO}}
<
variable sum sum2 num
constructor {} {
Line 2,541 ⟶ 4,299:
set sd [$sdacc value $val]
}
puts "the standard deviation is: $sd"</
{{out}}
<pre>the standard deviation is: 2.0</pre>
===With a Coroutine===
{{works with|Tcl|8.6}}
<
coroutine sd apply {{} {
set sum 0.0
Line 2,566 ⟶ 4,324:
}
sd stop
puts "the standard deviation is: $sd"</
[[Category:Stateful transactions]]
=={{header|TI-83 BASIC}}==
On the TI-83 family, standard deviation of a population is
a builtin function (σx):
• Press [STAT] select [EDIT] followed by [ENTER]
• then enter for list L1 in the table : 2, 4, 4, 4, 5, 5, 7, 9
• Or enter {2,4,4,4,5,5,7,9}→L1
• Press [STAT] select [CALC] then [1-Var Stats] select list L1 followed by [ENTER]
• Then σx (=2) gives the standard deviation of the population
=={{header|VBScript}}==
<syntaxhighlight lang="vb">data = Array(2,4,4,4,5,5,7,9)
For i = 0 To UBound(data)
WScript.StdOut.Write "value = " & data(i) &_
" running sd = " & sd(data,i)
WScript.StdOut.WriteLine
Next
Function sd(arr,n)
mean = 0
variance = 0
For j = 0 To n
mean = mean + arr(j)
Next
mean = mean/(n+1)
For k = 0 To n
variance = variance + ((arr(k)-mean)^2)
Next
variance = variance/(n+1)
sd = FormatNumber(Sqr(variance),6)
End Function</syntaxhighlight>
{{Out}}
<pre>
value = 2 running sd = 0.000000
value = 4 running sd = 1.000000
value = 4 running sd = 0.942809
value = 4 running sd = 0.866025
value = 5 running sd = 0.979796
value = 5 running sd = 1.000000
value = 7 running sd = 1.399708
value = 9 running sd = 2.000000
</pre>
=={{header|Visual Basic}}==
Line 2,574 ⟶ 4,376:
Note that the helper function <code>avg</code> is not named <code>average</code> to avoid a name conflict with <code>WorksheetFunction.Average</code> in MS Excel.
<
'treats non-numeric strings as zero
Dim L0 As Variant, total As Variant
Line 2,623 ⟶ 4,425:
Debug.Print standardDeviation(x(L0))
Next
End Sub</
{{out}}
<pre>
0
1
Line 2,634 ⟶ 4,437:
1.39970842444753
2
</pre>
=={{header|Wren}}==
{{libheader|Wren-fmt}}
{{libheader|Wren-math}}
<syntaxhighlight lang="wren">import "./fmt" for Fmt
import "./math" for Nums
var cumStdDev = Fiber.new { |a|
for (i in 0...a.count) {
var b = a[0..i]
System.print("Values : %(b)")
Fiber.yield(Nums.popStdDev(b))
}
}
var a = [2, 4, 4, 4, 5, 5, 7, 9]
while (true) {
var sd = cumStdDev.call(a)
if (cumStdDev.isDone) return
Fmt.print("Std Dev : $10.8f\n", sd)
}</syntaxhighlight>
{{out}}
<pre>
Values : [2]
Std Dev : 0.00000000
Values : [2, 4]
Std Dev : 1.00000000
Values : [2, 4, 4]
Std Dev : 0.94280904
Values : [2, 4, 4, 4]
Std Dev : 0.86602540
Values : [2, 4, 4, 4, 5]
Std Dev : 0.97979590
Values : [2, 4, 4, 4, 5, 5]
Std Dev : 1.00000000
Values : [2, 4, 4, 4, 5, 5, 7]
Std Dev : 1.39970842
Values : [2, 4, 4, 4, 5, 5, 7, 9]
Std Dev : 2.00000000
</pre>
=={{header|XPL0}}==
<
int A, I;
real N, S, S2;
Line 2,648 ⟶ 4,500:
];
CrLf(0);
]</
{{out}}
<pre>
0.00000 1.00000 0.94281 0.86603 0.97980 1.00000 1.39971 2.00000
Line 2,656 ⟶ 4,508:
=={{header|zkl}}==
<
m:=xs.append(x.toFloat()).sum(0.0)/xs.len();
(xs.reduce('wrap(p,x){(x-m)*(x-m) +p},0.0)/xs.len()).sqrt()
}.fp1(L())
}</
{{out}}
<pre>
zkl: T(2,4,4,4,5,5,7,9).pump(Void,sdf())
|