Differences

This shows you the differences between two versions of the page.

--- r:sampling_distribution [2025/09/10 14:02] – [Last one . . . Important] hkimscil
+++ r:sampling_distribution [2025/09/10 20:41] (current) – [PS1. week02] hkimscil
@@ Line 1: / Line 1: @@
 ====== PS1. week02 ======
 <code>
 rm(list=ls())
@@ Line 32: / Line 31: @@
 var(p1)
-hist(p1, breaks=100, col=rgb(1,1,1,1))
+hist(p1, breaks=50, col = rgb(1, 1, 1, 0.5),
+     main = "histogram of p1 and p2",)
+abline(v=mean(p1), col="black", lwd=3)
+hist(p2, add=T, breaks=50, col=rgb(1,1,.5,.5))
+abline(v=mean(p2), col="red", lwd=3)
+hist(p1, breaks=50, col=rgb(0,.5,.5,.5))
 abline(v=mean(p1),lwd=2)
 abline(v=mean(p1)-sd(p1), lwd=2)
@@ Line 90: / Line 97: @@
 pnorm(1.8)-pnorm(-1.8)
-hist(z.p1, breaks=100, col=rgb(0,0,0,0))
+hist(z.p1, breaks=50, col=rgb(1,0,0,0))
 abline(v=c(m.p1, -1.8, 1.8), col='red')
 -(pnorm(1.8)-pnorm(-1.8))
@@ Line 104: / Line 111: @@
 #
-hist(p1, breaks=100, col=rgb(1,1,1,1))
+hist(p1, breaks=50, col=rgb(.9,.9,.9,.9))
 abline(v=mean(p1),lwd=2)
 abline(v=mean(p1)-sd(p1), lwd=2)
@@ Line 116: / Line 123: @@
 c(a, b)
 c(-1, 1)
+# note that
+.32/2
+pnorm(-1)
+qnorm(.32/2)
+qnorm(pnorm(-1))
 # 95%
@@ Line 122: / Line 134: @@
 c(c, d)
 c(-2,2)
 # 99%
 e <- qnorm(.01/2)
@@ Line 127: / Line 140: @@
 c(e,f)
 c(-3,3)
 pnorm(b)-pnorm(a)
@@ Line 140: / Line 154: @@
 ################################
-hist(p1, breaks=50, col = rgb(1, 0, 0, 0.5),
-     main = "histogram of p1 and p2",)
-abline(v=mean(p1), col="black", lwd=3)
-hist(p2, add=T, breaks=50, col=rgb(0,0,1,.5))
-abline(v=mean(p2), col="violet", lwd=3)
 s.size <- 10
@@ Line 170: / Line 178: @@
 se.s <- sd(means)
-hist(means, breaks=100, col=rgb(.1, 0, 0, .5))
+hist(means, breaks=50,
-abline(v=mean(means), col="red", lwd=2)
+     xlim = c(mean(means)-5*sd(means), mean(means)+10*sd(means)),
+     col=rgb(1, 1, 1, .5))
+abline(v=mean(means), col="black", lwd=3)
 # now we want to get sd of this distribution
 lo1 <- mean(means)-se.s
@@ Line 180: / Line 189: @@
 lo3 <- mean(means)-3*se.s
 hi3 <- mean(means)+3*se.s
-hist(means,
-     xlim = c(mean(means)-5*sd(means), mean(means)+10*sd(means)),
-     col = rgb(1, 0, 0, .5))
 abline(v=mean(means), col="black", lwd=2)
-# abline(v=mean(p2), colo='darkgreen', lwd=3)
+# abline(v=mean(p2), colo='darkgreen', lwd=2)
 abline(v=c(lo1, hi1, lo2, hi2, lo3, hi3),
-       col=c("green","green", "blue", "blue", "orange", "orange"),
+       col=c("red","red", "blue", "blue", "orange", "orange"),
        lwd=2)
@@ Line 198: / Line 203: @@
 # sd of sample means (sd(means))
-# is sqrt(var(s1)/s.size) or
-# sd(s1) / sqrt(s.size)
 # = se.s
 # when iter value goes to
-# unlimited value:
+# infinite value:
 # mean(means) = mean(p1)
 # and
 # sd(means) = sd(p1) / sqrt(s.size)
-# that is, sd(means) = se.z
+# that is, se.s = se.z
 # This is called CLT (Central Limit Theorem)
+# see http://commres.net/wiki/cetral_limit_theorem
 mean(means)
 mean(p1)
 sd(means)
 var(p1)
+# remember we started talking sample size 10
 sqrt(var(p1)/s.size)
 se.z
@@ Line 237: / Line 243: @@
-hist(means,
+hist(means, breaks=50,
      xlim = c(mean(means)-5*sd(means), mean(means)+10*sd(means)),
-     col = rgb(1, 0, 0, .5))
+     col = rgb(1, 1, 1, .5))
-abline(v=mean(means), col="black", lwd=2)
+abline(v=mean(means), col="black", lwd=3)
 # abline(v=mean(p2), colo='darkgreen', lwd=3)
 abline(v=c(lo1, hi1, lo2, hi2, lo3, hi3),
-       col=c("green","green", "blue", "blue", "orange", "orange"),
+       col=c("darkgreen","darkgreen", "blue", "blue", "orange", "orange"),
        lwd=2)
@@ Line 257: / Line 263: @@
 m.sample.i.got
-hist(means,
+hist(means, breaks=30,
-     xlim = c(mean(means)-10*sd(means), mean(means)+10*sd(means)),
+     xlim = c(mean(means)-7*sd(means), mean(means)+10*sd(means)),
-     col = rgb(1, 0, 0, .5))
+     col = rgb(1, 1, 1, .5))
 abline(v=mean(means), col="black", lwd=3)
 abline(v=m.sample.i.got, col='darkgreen', lwd=3)
@@ Line 276: / Line 282: @@
 # (green line)
 tmp <- mean(means) - (m.sample.i.got - mean(means))
-abline(v=tmp, col='green', lwd=3)
+abline(v=tmp, col='red', lwd=3)
 * pnorm(m.sample.i.got, mean(p1), sd(means), lower.tail = F)
 m.sample.i.got
 ### one more time
+# this time, with a story
 mean(p2)
 sd(p2)
@@ Line 287: / Line 294: @@
 m.sample.i.got
-hist(means,
+tmp <- mean(means) - (m.sample.i.got-mean(means))
-     xlim = c(mean(means)-15*sd(means), mean(means)+15*sd(means)),
+tmp
-     col = rgb(1, 0, 0, .5))
-abline(v=mean(means), col="black", lwd=2)
+hist(means, breaks=30,
-abline(v=m.sample.i.got, col='darkgreen', lwd=2)
+     xlim = c(tmp-4*sd(means), m.sample.i.got+4*sd(means)),
+     col = rgb(1, 1, 1, .5))
+abline(v=mean(means), col="black", lwd=3)
+abline(v=m.sample.i.got, col='blue', lwd=3)
 # what is the probablity of getting
@@ Line 304: / Line 314: @@
 # mean(means) - m.sample.i.got - mean(means)
 # (green line)
-tmp <- mean(means) - (m.sample.i.got - mean(means))
+abline(v=tmp, col='red', lwd=3)
-abline(v=tmp, col='green', lwd=2)
 * pnorm(m.sample.i.got, mean(p1), sd(means), lower.tail = F)
 </code>
 ====== output ======
@@ Line 644: / Line 649: @@
 </WRAP>
 <WRAP column half>
-....
+qnorm
   * qnorm는 pnorm의 반대값을 구하는 명령어
   * 히스토그램에서 검정 색 부분의 바깥 쪽 부분은 32%이고 왼 쪽의 것은 이것의 반인 16% 이다.