——————————————————————————————————————————————————————————–
Base de dados: PENDRIVE (N:)
## [1] "N:/_IFMG/_MESTRADO/ANALISE_DE_DADOS/trabalho_01"
Apaga variáveis do ambiente R:
rm(list=ls())
Carregando bibliotecas R:
Biblioteca PWR:
#install.packages("pwr")
library(pwr)
Biblioteca para Tabela de Distribuição de Frequências:
#install.packages("sm") # comando para instalação da biblioteca sm #
library("sm")
——————————————————————————————————————————————————————————–
Lista de Exercícios - Tema: Intervalos de Confiança e Testes de Hipóteses
Regras:
Nível de confiança: 5% (0.95)
Desenvolvimento:
vet_obs<-c(57,60,49,50,51,60,49,53,49,56,64,60,49,52,69,40,44,38,53,66)
summary(vet_obs)
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 38.00 49.00 52.50 53.45 60.00 69.00
stem(sort(vet_obs))
##
## The decimal point is 1 digit(s) to the right of the |
##
## 3 | 8
## 4 | 049999
## 5 | 0123367
## 6 | 000469
# Análise via boxplot:
boxplot(vet_obs)
# Aplicando t.test:
#Desvio padrão das observações:
sd(vet_obs)
## [1] 8.165879
# Teste com alpha=5% Opção: two.sided - mu=0
t.test(vet_obs,alternative=c("two.sided"),mu=50,conf.level=0.95)
##
## One Sample t-test
##
## data: vet_obs
## t = 1.8894, df = 19, p-value = 0.07419
## alternative hypothesis: true mean is not equal to 50
## 95 percent confidence interval:
## 49.62825 57.27175
## sample estimates:
## mean of x
## 53.45
# Cálculo da amplitude do IC:
r95<-t.test(vet_obs,alternative=c("two.sided"),mu=50,conf.level=0.95)
r95$conf.int[2]-r95$conf.int[1]
## [1] 7.643498
# Teste de Wilcox
wilcox.test(vet_obs,mu=50,conf.int=TRUE)
##
## Wilcoxon signed rank test with continuity correction
##
## data: vet_obs
## V = 139, p-value = 0.07905
## alternative hypothesis: true location is not equal to 50
## 95 percent confidence interval:
## 49.00001 58.00001
## sample estimates:
## (pseudo)median
## 53.66572
CONCLUSÃO:
Para as 20 observações analisadas e dentro da interpretação de conformidade da indústria, observa-se que:
Conclui-se: Apesar do valor em questao encontrar-se dentro do intervalo de confiança e em conformidade com o padrão estabelecido, a amplitude superior a 7.64 mg/l, desvio padrão superior a 8 mg/l e a proximidade com o limite inferior do IC, recomenda-se a NÃO ACEITAÇÃO DE CONFORMIDADE baseando-se nas observações analisadas.
——————————————————————————————————————————————————————————–
Regras:
O Viés será estimado pela diferença do valor referência (50) dentro do IC.
Amostra: c(50.3, 51.2, 50.5, 50.2, 49.9, 50.2, 50.3, 50.5, 49.3, 50.0, 50.4, 50.1, 51.0, 49.8, 50.7, 50.6)
Desenvolvimento:
Amostras<-c(50.3, 51.2, 50.5, 50.2, 49.9, 50.2, 50.3, 50.5, 49.3, 50.0, 50.4, 50.1, 51.0, 49.8, 50.7, 50.6)
summary(Amostras)
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 49.30 50.08 50.30 50.31 50.52 51.20
stem(sort(Amostras))
##
## The decimal point is at the |
##
## 49 | 3
## 49 | 89
## 50 | 0122334
## 50 | 5567
## 51 | 02
# Análise via boxplot:
boxplot(Amostras)
# Desvio padrão das observações (n=16):
sd(Amostras)
## [1] 0.4616998
# Cálculo do viés:
# t=1.96 (alfa=5%)
# n=16
Vies<-1.96*sd(Amostras)/sqrt(16)
Vies
## [1] 0.2262329
# Aplicando t.test para análise da hipótese:
# Teste com alpha=5% Opção: two.sided
t.test(Amostras,alternative=c("two.sided"),mu=50,conf.level=0.95)
##
## One Sample t-test
##
## data: Amostras
## t = 2.7074, df = 15, p-value = 0.01622
## alternative hypothesis: true mean is not equal to 50
## 95 percent confidence interval:
## 50.06648 50.55852
## sample estimates:
## mean of x
## 50.3125
r95<-t.test(Amostras,alternative=c("two.sided"),mu=50,conf.level=0.95)
r95$conf.int[2]-r95$conf.int[1]
## [1] 0.4920449
# Teste com alpha=1% Opção: two.sided
t.test(Amostras,alternative=c("two.sided"),mu=0,conf.level=0.99)
##
## One Sample t-test
##
## data: Amostras
## t = 435.8893, df = 15, p-value < 2.2e-16
## alternative hypothesis: true mean is not equal to 0
## 99 percent confidence interval:
## 49.97238 50.65262
## sample estimates:
## mean of x
## 50.3125
r99<-t.test(Amostras,alternative=c("two.sided"),mu=0,conf.level=0.99)
r99$conf.int[2]-r99$conf.int[1]
## [1] 0.6802483
CONCLUSÃO:
Após a análise dos dados, conclui-se:
——————————————————————————————————————————————————————————–
Regras:
n=7 (64,65,75,67,65,74,75)
Amostras<-c(64,65,75,67,65,74,75)
summary(Amostras)
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 64.00 65.00 67.00 69.29 74.50 75.00
# Análise via boxplot:
boxplot(Amostras)
# Desvio padrão das observações (n=7):
sd(Amostras)
## [1] 5.122313
# Cálculo (considerando alfa=10%):
# t=1.645
# n=7
Dif<-1.645*sd(Amostras)/sqrt(7)
Dif
## [1] 3.184806
# Teste com alpha=10% Opção: two.sided
t.test(Amostras,alternative=c("two.sided"),mu=72,conf.level=0.90)
##
## One Sample t-test
##
## data: Amostras
## t = -1.402, df = 6, p-value = 0.2105
## alternative hypothesis: true mean is not equal to 72
## 90 percent confidence interval:
## 65.52362 73.04781
## sample estimates:
## mean of x
## 69.28571
r10<-t.test(Amostras,alternative=c("two.sided"),mu=72,conf.level=0.90)
r10$conf.int[2]-r10$conf.int[1]
## [1] 7.524198
# Cálculo (considerando alfa=5%):
# t=1.96
# n=7
Dif<-1.96*sd(Amostras)/sqrt(7)
Dif
## [1] 3.794663
# Teste com alpha=5% Opção: two.sided
t.test(Amostras,alternative=c("two.sided"),mu=72,conf.level=0.95)
##
## One Sample t-test
##
## data: Amostras
## t = -1.402, df = 6, p-value = 0.2105
## alternative hypothesis: true mean is not equal to 72
## 95 percent confidence interval:
## 64.54836 74.02306
## sample estimates:
## mean of x
## 69.28571
r5<-t.test(Amostras,alternative=c("two.sided"),mu=72,conf.level=0.95)
r5$conf.int[2]-r5$conf.int[1]
## [1] 9.4747
# Cálculo (considerando alfa=1%):
# t=2.58
# n=7
Dif<-2.58*sd(Amostras)/sqrt(7)
Dif
## [1] 4.995015
# Teste com alpha=1% Opção: two.sided
t.test(Amostras,alternative=c("two.sided"),mu=72,conf.level=0.99)
##
## One Sample t-test
##
## data: Amostras
## t = -1.402, df = 6, p-value = 0.2105
## alternative hypothesis: true mean is not equal to 72
## 99 percent confidence interval:
## 62.10794 76.46349
## sample estimates:
## mean of x
## 69.28571
r1<-t.test(Amostras,alternative=c("two.sided"),mu=72,conf.level=0.99)
r1$conf.int[2]-r1$conf.int[1]
## [1] 14.35555
CONCLUSÃO:
Após a análise dos dados, conclui-se:
——————————————————————————————————————————————————————————–
Regras:
n=9 (igual para as duas amostras)
AmISE<-c(0.32,0.36,0.24,0.11,0.11,0.44,2.79,2.99,3.47)
AmCOL<-c(0.36,0.37,0.21,0.09,0.11,0.42,2.77,2.91,3.52)
summary(AmISE)
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.110 0.240 0.360 1.203 2.790 3.470
summary(AmCOL)
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.090 0.210 0.370 1.196 2.770 3.520
# Análise via boxplot:
par(mfrow=c(1,2))
boxplot(AmISE,main="Amostra ISE")
boxplot(AmCOL,main="Amostra COLORIMETRIC")
Análise da amostra ISE:
# Desvio padrão das observações (n=9):
sd(AmISE)
## [1] 1.424798
# Cálculo (considerando alfa=5%):
# t=1.96
# n=9
Dif1<-1.96*sd(AmISE)/sqrt(9)
Dif1
## [1] 0.9308682
# Teste com alpha=5% Opção: two.sided
t.test(AmISE,alternative=c("two.sided"),mu=0,conf.level=0.95)
##
## One Sample t-test
##
## data: AmISE
## t = 2.5337, df = 8, p-value = 0.03505
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 0.1081365 2.2985302
## sample estimates:
## mean of x
## 1.203333
rAmISE5<-t.test(AmISE,alternative=c("two.sided"),mu=0,conf.level=0.95)
rAmISE5$conf.int[2]-rAmISE5$conf.int[1]
## [1] 2.190394
Análise da amostra COLORIMETRIC:
# Desvio padrão das observações (n=9):
sd(AmCOL)
## [1] 1.421901
# Cálculo (considerando alfa=5%):
# t=1.96
# n=9
Dif2<-1.96*sd(AmCOL)/sqrt(9)
Dif2
## [1] 0.9289754
# Teste com alpha=5% Opção: two.sided
t.test(AmCOL,alternative=c("two.sided"),mu=0,conf.level=0.95)
##
## One Sample t-test
##
## data: AmCOL
## t = 2.5224, df = 8, p-value = 0.03567
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 0.1025856 2.2885255
## sample estimates:
## mean of x
## 1.195556
rAmCOL5<-t.test(AmCOL,alternative=c("two.sided"),mu=0,conf.level=0.95)
rAmCOL5$conf.int[2]-rAmCOL5$conf.int[1]
## [1] 2.18594
#Cálculo t.test para as duas amostras:
# Teste com alpha=5% Opção: two.sided
t.test(AmCOL,AmISE,alternative=c("two.sided"),mu=0,conf.level=0.95)
##
## Welch Two Sample t-test
##
## data: AmCOL and AmISE
## t = -0.0116, df = 16, p-value = 0.9909
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -1.430179 1.414624
## sample estimates:
## mean of x mean of y
## 1.195556 1.203333
CONCLUSÃO:
Após a análise das duas amostras ISE e COLORIMETRIC, conclui-se:
——————————————————————————————————————————————————————————–
Regras:
nAmPAR=10
AmCID<-c(0.34,0.18,0.13,0.09,0.16,0.09,0.16,0.10,0.14,0.26,0.06,0.26,0.07)
AmPAR<-c(0.26,0.06,0.16,0.19,0.32,0.16,0.08,0.05,0.10,0.13)
summary(AmCID)
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.0600 0.0900 0.1400 0.1569 0.1800 0.3400
summary(AmPAR)
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.0500 0.0850 0.1450 0.1510 0.1825 0.3200
Análise via diagrama CAULE/FOLHAS:
stem(sort(AmCID))
##
## The decimal point is 1 digit(s) to the left of the |
##
## 0 | 6799
## 1 | 034668
## 2 | 66
## 3 | 4
stem(sort(AmPAR))
##
## The decimal point is 1 digit(s) to the left of the |
##
## 0 | 568
## 1 | 03669
## 2 | 6
## 3 | 2
Análise via boxplot:
par(mfrow=c(1,2))
boxplot(AmCID,main="Amostra CIDADES")
boxplot(AmPAR,main="Amostra poços PARTICULARes")
Análise da amostra CIDade:
# Desvio padrão das observações (n=13):
sd(AmCID)
## [1] 0.08439832
# Cálculo (considerando alfa=5%):
# t=1.96
# n=13
Dif1<-1.96*sd(AmCID)/sqrt(13)
Dif1
## [1] 0.04587945
# Teste com alpha=5% Opção: two.sided
t.test(AmCID,alternative=c("two.sided"),mu=0,conf.level=0.95)
##
## One Sample t-test
##
## data: AmCID
## t = 6.7039, df = 12, p-value = 2.184e-05
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 0.1059217 0.2079245
## sample estimates:
## mean of x
## 0.1569231
rAmCID5<-t.test(AmCID,alternative=c("two.sided"),mu=0,conf.level=0.95)
rAmCID5$conf.int[2]-rAmCID5$conf.int[1]
## [1] 0.1020028
Análise da amostra PARticular:
# Desvio padrão das observações (n=10):
sd(AmPAR)
## [1] 0.08736259
# Cálculo (considerando alfa=5%):
# t=1.96
# n=10
Dif2<-1.96*sd(AmPAR)/sqrt(10)
Dif2
## [1] 0.05414789
# Teste com alpha=5% Opção: two.sided
t.test(AmPAR,alternative=c("two.sided"),mu=0,conf.level=0.95)
##
## One Sample t-test
##
## data: AmPAR
## t = 5.4658, df = 9, p-value = 0.0003974
## alternative hypothesis: true mean is not equal to 0
## 95 percent confidence interval:
## 0.08850457 0.21349543
## sample estimates:
## mean of x
## 0.151
rAmPAR5<-t.test(AmPAR,alternative=c("two.sided"),mu=0,conf.level=0.95)
rAmPAR5$conf.int[2]-rAmPAR5$conf.int[1]
## [1] 0.1249909
#Cálculo t.test para as duas amostras:
# Teste com alpha=5% Opção: two.sided
t.test(AmPAR,AmCID,alternative=c("two.sided"),mu=0,conf.level=0.95)
##
## Welch Two Sample t-test
##
## data: AmPAR and AmCID
## t = -0.1636, df = 19.156, p-value = 0.8718
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -0.08166929 0.06982314
## sample estimates:
## mean of x mean of y
## 0.1510000 0.1569231
CONCLUSÃO:
Após a análise das duas amostras CIDade e PARticular, conclui-se:
——————————————————————————————————————————————————————————–