joonised
Histogramm
ounad=read.table("http://www.tlu.ee/~jaagup/andmed/muu/ounad/antoonovka2.txt",
sep=",", header=TRUE)
head(ounad)## august september
## 1 6.0 7.9
## 2 4.0 5.7
## 3 5.2 6.6
## 4 4.1 5.4
## 5 5.7 7.9
## 6 4.2 5.9 hist(ounad$august, yaxp=c(0, 25, 5), ylim=c(0, 25), las=1, xlab="diameeter", ylab="kogus",
main="Õunte läbimõõtude jaotus")
abline(v=mean(ounad$august), col="red")
text(mean(ounad$august)-0.25, 10, "keskmine", srt=90)
points(3, 15, col="green")
points(5:7, c(15, 15, 15), col="green")
segments(2, 0, 5, 20)
# lines(density(ounad$august))Tihedusgraafik
plot(density(ounad$august), xlim=c(0, 10))
lines(density(ounad$september), col="blue")
XY
plot(ounad$august, ounad$september)
#Lisage joonisele seletused
plot(ounad$august, ounad$september, xlab="Diameeter augustis, cm", ylab="Diameeter septembris, cm",
main="Õunte läbimõõdud")
paljukasvanud=ounad[ ounad$september/ounad$august>1.4, ]
points(paljukasvanud$august, paljukasvanud$september, col="red")
#Värvige siniseks õunad, mis on kasvanud vähem kui 1.5 cm
vahekasvanud=ounad[ounad$september-ounad$august<1.5,]
points(vahekasvanud$august, vahekasvanud$september, col="blue")
Jitter
Muuda algandmeid vaevumärgatavalt. Praegu selleks, et samade koordinaatidega õunad ei satuks üksteise peale
plot(jitter(ounad$august), jitter(ounad$september),
col=rgb(0, 0, 0, 0.3), pch=19, cex=3)
Karp ja vurrud
boxplot(ounad$august)
#install.packages("reshape")
library(reshape)
pikk=melt(ounad)## Using as id variables head(pikk)## variable value
## 1 august 6.0
## 2 august 4.0
## 3 august 5.2
## 4 august 4.1
## 5 august 5.7
## 6 august 4.2 tail(pikk)## variable value
## 195 september 6.1
## 196 september 6.3
## 197 september 5.1
## 198 september 5.0
## 199 september 7.9
## 200 september 8.4 colnames(pikk)=c("kuu", "diameeter")
head(pikk)## kuu diameeter
## 1 august 6.0
## 2 august 4.0
## 3 august 5.2
## 4 august 4.1
## 5 august 5.7
## 6 august 4.2 boxplot(pikk$diameeter~pikk$kuu)
#Joonistage histogramm ning karp-ja-vurrud diagramm
#juurdekasvude absoluutse ja suhtelise jaotuse kohta
vahesentimeetrid=ounad$september-ounad$august
suhe=ounad$september/ounad$august
hist(vahesentimeetrid)
boxplot(vahesentimeetrid)
hist(suhe)
boxplot(suhe)
head(ounad)## august september
## 1 6.0 7.9
## 2 4.0 5.7
## 3 5.2 6.6
## 4 4.1 5.4
## 5 5.7 7.9
## 6 4.2 5.9 quantile(ounad$august)## 0% 25% 50% 75% 100%
## 2.100 4.175 4.850 5.500 7.000 plot(quantile(ounad$august))
plot(quantile(ounad$august), xaxt="n")
mtext(c("0%", "25%", "50%", "75%", "100%"),
side=1, at=1:5)
kvantiilid=quantile(ounad$august, seq(0, 1, 0.1))
plot(kvantiilid, xaxt="n")
mtext(names(kvantiilid), side=1, at=1:length(kvantiilid), cex=0.5)
plot(ounad$august, ounad$september) #iga täpp on sama õun kahel kuul
qqnorm(ounad$august)
qqnorm(ounad$september)
qqline(ounad$september)
qqplot(ounad$august, ounad$september)
Kahe õunasordi võrdlemine
ounad=read.table("http://www.tlu.ee/~jaagup/andmed/muu/ounad/ounad1000.txt", header=TRUE, sep=",")
mean(ounad$diameeter)## [1] 3.75649 quantile(ounad$diameeter)## 0% 25% 50% 75% 100%
## 0.0100 2.4575 3.5650 4.8225 10.1000 quantile(ounad$diameeter, seq(0, 1, 0.1))## 0% 10% 20% 30% 40% 50% 60% 70% 80% 90%
## 0.010 1.549 2.228 2.727 3.150 3.565 4.024 4.590 5.260 6.222
## 100%
## 10.100 #Kuvage õunte diameetrite detsiilid joonisel
#võimalusel lisage ka protsendisildid
kvantiilid=quantile(ounad$diameeter, seq(0, 1, 0.1))
plot(kvantiilid, xaxt="n")
mtext(names(kvantiilid), side=1, at=1:length(kvantiilid), cex=0.5)
mean(ounad$diameeter)## [1] 3.75649 head(ounad[ounad$ounasort=="Liivi sibul", ])## ounasort diameeter
## 1 Liivi sibul 2.60
## 2 Liivi sibul 2.43
## 3 Liivi sibul 3.67
## 5 Liivi sibul 3.33
## 6 Liivi sibul 3.72
## 10 Liivi sibul 2.53 ounad[ounad$ounasort=="Liivi sibul", "diameeter"]## [1] 2.60 2.43 3.67 3.33 3.72 2.53 2.49 2.38 2.78 0.38 3.66 2.71 4.76 3.46
## [15] 3.16 3.76 4.53 1.46 2.76 4.46 3.20 3.23 3.43 3.72 1.46 2.86 4.88 3.85
## [29] 3.86 2.98 4.49 3.59 1.77 1.99 3.07 1.17 4.00 3.43 1.91 2.66 1.75 3.97
## [43] 2.73 3.56 4.21 3.26 0.01 3.03 4.95 2.25 3.53 2.32 2.71 2.19 3.13 3.57
## [57] 4.50 3.03 3.78 6.60 2.44 3.28 2.93 4.52 3.44 4.41 2.96 2.58 3.07 1.25
## [71] 2.89 2.31 1.75 1.91 2.41 1.83 3.09 4.04 1.36 4.19 2.67 2.02 2.69 1.81
## [85] 2.39 2.98 4.75 3.25 1.67 3.43 3.23 3.17 4.21 1.51 3.69 2.45 3.22 4.68
## [99] 4.60 3.69 4.04 2.82 3.45 1.93 5.02 3.25 1.58 3.99 2.33 3.78 2.69 5.00
## [113] 4.80 1.68 3.22 4.29 1.71 2.33 3.15 3.19 2.85 3.64 1.98 2.50 3.10 3.74
## [127] 3.11 2.06 4.35 2.28 4.75 1.35 3.13 2.98 2.10 2.59 4.47 3.17 3.46 3.46
## [141] 4.22 3.72 3.04 3.11 0.66 3.76 1.75 3.27 4.34 2.96 3.81 1.86 4.64 2.60
## [155] 4.76 4.14 3.01 1.70 1.54 3.79 3.61 3.48 3.18 3.35 0.83 3.10 4.76 2.46
## [169] 3.37 4.99 2.27 4.16 2.08 2.30 1.78 2.80 3.05 3.17 2.26 4.20 2.59 2.96
## [183] 3.21 2.24 2.39 2.46 1.52 2.73 2.21 2.41 3.81 3.50 1.98 3.04 2.33 3.28
## [197] 3.39 2.62 3.34 4.13 3.51 0.53 3.72 4.70 2.96 2.62 4.60 3.97 2.44 3.63
## [211] 3.69 1.65 3.13 3.33 4.37 4.07 2.47 2.29 3.43 3.89 3.61 0.59 2.28 4.67
## [225] 4.55 2.18 2.83 2.71 3.19 3.30 2.09 2.82 3.78 2.62 1.75 5.22 3.07 2.94
## [239] 3.16 2.73 2.42 3.69 0.45 2.11 3.74 3.17 1.20 2.42 1.83 4.80 2.94 2.89
## [253] 3.64 1.32 3.83 2.33 2.44 2.41 3.24 3.76 2.88 1.91 2.27 3.89 2.13 1.73
## [267] 2.57 1.97 2.51 3.23 3.15 4.60 2.67 4.80 1.82 4.50 4.00 1.99 2.61 3.67
## [281] 1.59 2.78 2.50 1.03 2.93 2.98 1.37 2.86 1.76 mean(ounad$diameeter)## [1] 3.75649 mean(ounad[ounad$ounasort=="Liivi sibul", "diameeter"])## [1] 3.011038 mean(ounad[ounad$ounasort=="Kuldrenett", "diameeter"])## [1] 4.059494 tapply(ounad$diameeter, ounad$ounasort, mean)## Kuldrenett Liivi sibul
## 4.059494 3.011038 tapply(ounad$diameeter, ounad$ounasort, max)## Kuldrenett Liivi sibul
## 10.1 6.6 tapply(ounad$diameeter, ounad$ounasort, min)## Kuldrenett Liivi sibul
## 0.01 0.01 tapply(ounad$diameeter, ounad$ounasort,
function(diameetrid){max(diameetrid)-min(diameetrid)})## Kuldrenett Liivi sibul
## 10.09 6.59 barplot(tapply(ounad$diameeter, ounad$ounasort, mean))
barplot(tapply(ounad$diameeter, ounad$ounasort, mean),
main="Diameetrite võrdlus",
ylab="Keskmine diameeter (cm)")
boxplot(ounad$diameeter~ounad$ounasort)
t.test(ounad$diameeter)##
## One Sample t-test
##
## data: ounad$diameeter
## t = 65.34, df = 999, p-value < 2.2e-16
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 3.643672 3.869308
## sample estimates:
## mean of x
## 3.75649Nende õunte diameetrite põhjal võin väita, et 95% tõenäosusega on selle aia õunte diameetrite keskmine vahemikus 3,64 kuni 3,86 cm
Ehk siis: kui korjan korduvalt 1000 õuna ja mõõdan diameetrid, siis keskmiselt 95-l juhul sajast jääb nende mõõdetud diameetrite keskmine nimetatud vahemikku
95 percent confidence interval: 3.643672 3.869308
99 percent confidence interval: 3.608118 3.904862
t.test(ounad$diameeter, conf.level = 0.99)##
## One Sample t-test
##
## data: ounad$diameeter
## t = 65.34, df = 999, p-value < 2.2e-16
## alternative hypothesis: true mean is not equal to 0
## 99 percent confidence interval:
## 3.608118 3.904862
## sample estimates:
## mean of x
## 3.75649Võrdlus eelmise aasta keskmisega (4.0)
t.test(ounad$diameeter, mu=4.0)##
## One Sample t-test
##
## data: ounad$diameeter
## t = -4.2356, df = 999, p-value = 2.49e-05
## alternative hypothesis: true mean is not equal to 4
## 95 percent confidence interval:
## 3.643672 3.869308
## sample estimates:
## mean of x
## 3.75649t = -4.2356, df = 999, p-value = 2.49e-05 Pragune keskmine (3,75) on võrreldavast keskmisest (4.0) 4,23 Studenti hälbe jagu väiksem Tõenäosus, et praeguste õunte keskmine võiks juhuslikult siiski olla 4, on 2.49e-05 0,0000249
t.test(ounad$diameeter, mu=3.8)##
## One Sample t-test
##
## data: ounad$diameeter
## t = -0.75681, df = 999, p-value = 0.4493
## alternative hypothesis: true mean is not equal to 3.8
## 95 percent confidence interval:
## 3.643672 3.869308
## sample estimates:
## mean of x
## 3.75649Tõenäosus, et õunte keskmine diameeter võib olla 3,8 on 44,9%
t.test(ounad[ounad$ounasort=="Liivi sibul", "diameeter"],
ounad[ounad$ounasort=="Kuldrenett", "diameeter"])##
## Welch Two Sample t-test
##
## data: ounad[ounad$ounasort == "Liivi sibul", "diameeter"] and ounad[ounad$ounasort == "Kuldrenett", "diameeter"]
## t = -11, df = 946.53, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -1.2355042 -0.8614071
## sample estimates:
## mean of x mean of y
## 3.011038 4.05949495 percent confidence interval: -1.2355042 -0.8614071
95% tõenäosusega on Liivi sibulate keskmine diameeter Kuldrenettide keskmisest diameetist selles aias väiksem 0,86 kuni 1,23 cm võrra
#hist(ounad$diameeter)
nahtav=0.4
keskmine=mean(ounad$diameeter)
xpiirid=c((1-nahtav)*keskmine, (1+nahtav)*keskmine)
plot(density(ounad$diameeter), xlim=xpiirid)
abline(v=mean(ounad$diameeter), col="red", lty=2)
tulemus=t.test(ounad$diameeter, conf.level=0.99)
abline(v=tulemus$conf.int, col="blue")
ounad=read.table("http://www.tlu.ee/~jaagup/andmed/muu/ounad/antoonovka2.txt", sep=",", header=TRUE)
head(ounad)## august september
## 1 6.0 7.9
## 2 4.0 5.7
## 3 5.2 6.6
## 4 4.1 5.4
## 5 5.7 7.9
## 6 4.2 5.9 plot(ounad$august, ounad$september)
cor(ounad$august, ounad$september)## [1] 0.892967 cor.test(ounad$august, ounad$september)##
## Pearson's product-moment correlation
##
## data: ounad$august and ounad$september
## t = 19.639, df = 98, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.8447060 0.9268248
## sample estimates:
## cor
## 0.892967 cor.test(ounad$august, ounad$september, conf.level=0.90)##
## Pearson's product-moment correlation
##
## data: ounad$august and ounad$september
## t = 19.639, df = 98, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 90 percent confidence interval:
## 0.8536274 0.9221775
## sample estimates:
## cor
## 0.892967 lm(ounad$september~ounad$august)##
## Call:
## lm(formula = ounad$september ~ ounad$august)
##
## Coefficients:
## (Intercept) ounad$august
## 1.85 1.01 abline(lm(ounad$september~ounad$august))
lm(ounad\(september~ounad\)august) Coefficients: (Intercept) ounad$august
1.85 1.01
Õunte ennustatav diameeter septembris =
1.85cm + 1.01 x diameeter augustis