Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.
|
|
Luis A. Afonso
Posts:
4,277
From:
LIsbon (Portugal)
Registered:
2/16/05
|
|
“KNOTS”, check sample uniformity
Posted:
Dec 17, 2012 5:57 PM
|
|
?KNOTS?, check sample uniformity
___n= sample size
__________n=20_____40______60_____80_____100___
_fractile___
__0.005____0.109___0.114___0.114___0.115___0.116__
__0.010____0.118___0.122___0.123___0.123___0.124__
__0.025____0.132___0.136___0.137___0.137___0.138__
__0.050____0.147___0.150___0.151___0.152___0.152__
__0.175____0.158___0.161___0.161___0.162___0.163__
__0.100____0.167___0.170___0.171___0.171___0.171__
__0.200____0.197___0.199___0.200___0.200___0.201__
__0.300____0.223___0.225___0.226___0.226___0.226__
__0.400____0.250___0.252___0.252___0.252___0.253__
__0.500____0.279___0.280___0.281___0.281___0.281__
__0.600____0.312___0.313___0.314___0.314___0.314__
__0.700____0.353___0.354___0.355___0.355___0.355__
__0.800____0.408___0.409___0.409___0.409___0.410__
__0.900____0.497___0.498___0.499___0.499___0.499__
__0.925____0.532___0.532___0.534___0.534___0.534__
__0.950____0.579___0.580___0.580___0.581___0.581__
__0.975____0.655___0.655___0.657___0.657___0.658__
__0.990____0.745___0.747___0.750___0.751___0.750__
__0.995____0.805___0.813___0.814___0.814___0.814__
___Confidence Intervals
__Significance_ 5%___________ 1%_______
__n=20___[0.132, 0.655]___[0.109, 0.805]
____40___[0.136, 0.655]___[0.114, 0.813]
____60___[0.137, 0.657]___[0.114, 0.814]
____80___[0.137, 0.657]___[0.115, 0.814]
___100___[0.138, 0.658]___[0.116, 0.814]
Amazingly the CI bounds for different n are similar and seem to not grow without limit (asymptotic?).
Luis A. Afonso
REM "KNOTS"
REM
CLS
PRINT "0_|__|__|...|__|_1"
PRINT " knots: 1/(2*n), 3/(2*n), ...";
PRINT ",(2*n-3)/(2*n),(2*n-1)/(2*n) "
PRINT "Statistics-->"
PRINT "--> SUM di (j=1, n) = SQR((y(j) - kn(j)) ^ 2 / n)"
INPUT " n= "; n
INPUT " all= "; ali
DIM kn(n), X(n), XX(n), y(n)
DIM w(8001)
DIM u(30)
REM
FOR i = 1 TO n: kn(i) = (2 * i - 1) / (n + n): NEXT i
RANDOMIZE TIMER
FOR rpt = 1 TO ali
LOCATE 6, 50
PRINT USING "##########"; ali - rpt
PRINT : PRINT
FOR u = 1 TO n: X(u) = RND: XX(u) = X(u): NEXT u
FOR u1 = 1 TO n
w = X(u1): p = 1
FOR u2 = 1 TO n
IF XX(u2) < w THEN p = p + 1
NEXT u2
y(p) = w
NEXT u1
w = 0
FOR t = 1 TO n
d = SQR((y(t) - kn(t)) ^ 2 / n)
w = w + d
NEXT t
w = INT(w * 1000 + .5)
w(w) = w(w) + 1
au = rpt / 10000: bu = INT(au)
IF au <> bu THEN GOTO 20
u(1) = .005: u(2) = .01
u(3) = .025: u(4) = .05: u(5) = .075: u(6) = .1
u(7) = .2: u(8) = .3: u(9) = .4
u(10) = .5: u(11) = .6: u(12) = .7: u(13) = .8: u(14) = .9
u(15) = 1 - u(5): u(16) = 1 - u(4): u(17) = 1 - u(3)
u(18) = 1 - u(2): u(19) = 1 - u(1)
REM
FOR y = 1 TO 19
IF y = 1 THEN GOTO 7
IF y / 2 = INT(y / 2) THEN PRINT
7 su = 0: u = u(y)
FOR tt = 0 TO 8000
su = su + w(tt) / rpt
IF su > u THEN GOTO 17
NEXT tt
17 PRINT USING " #.### #.### "; u(y); tt / 1000;
NEXT y
20 NEXT rpt
END
|
|
|
|
