还是我们前两次的测试数据：
library(ggplot2)
data <- read.table("week.data.txt",header=T,sep="\t")
head(data)
ggplot(data, aes(x = day, y = frequency, fill = week_n)) +
geom_bar(stat = "identity", color = "white",lwd = 1, show.legend = FALSE,width = 0.6)+
scale_x_continuous(breaks=seq(1,21,1),labels=data$week)+
theme(text = element_text(size = 15))+
geom_errorbar(aes(ymin=frequency-sad, ymax=frequency+sad),width=0.2,position=position_dodge(0.6),size=0.8)
主要通过geom_errorbar来添加误差线。比如我们先测试的是已知的误差列sad。
我们也可以自己计算不同week之间的均值和方差，然后画图。
data_new <- data %>%
group_by(week_n) %>%
mutate(mean = mean(frequency), group_max = max(frequency),var=mad(frequency) ) %>%
arrange(desc(mean))
data_new2 <- unique(data_new[,c(2,8,9,10)])
my_comparisons <- list( c("week1", "week2"), c("week1", "week3"), c("week2", "week3") )
ggplot(data_new2, aes(x = week_n, y = mean, fill = week_n)) +
geom_bar(stat = "identity", color = "white",lwd = 1, show.legend = F,width = 0.6)+
theme(text = element_text(size = 15))+
geom_errorbar(aes(ymin=mean-var, ymax=mean+var),width=0.2,position=position_dodge(0.6),size=0.8)+
stat_compare_means(comparisons = my_comparisons)+
stat_compare_means(label.y = 60,label.x = 1.5)
图中，我们通过geom_errorbar加入了误差bar，并且运用了前面的技巧，通过stat_compare_means做了两两之间的统计显著性检验。
