R新包:快速分组进行多重比较
学习R两年半，我们的小伙伴终于发布了人生第一个R语言包（MultiCom）。
library(rio)library(tidyverse)library(MultiCom)set.seed(123)# 生成演示数据dat <- expand.grid(Year = 2018:2021,Site = c("Beijing", "Shanghai", "Guangzhou", "Chengdu"),Hybrid = paste0("H", 1:4),Organ = c("Leaf", "Stem")) %>%mutate(Value = runif(n(), 10, 50) %>% round(1)) %>%slice(rep(1:n(), 4)) %>%mutate(Value = Value + rnorm(n(), 0, 5) %>% round(1),Value = pmax(Value, 0.1)) %>%sample_n(n()) %>%tibble::rownames_to_column("ID") %>%select(-ID)# 分组多重比较result <- multi_group_com(data = dat,var_names = c("Hybrid", "Value"),group_vars = c("Site", "Year", "Organ"),method = c("LSD", 0.05))head(result)# 堆积柱状图数据格式转换result_plot <- result %>%mutate(Organ = factor(Organ, levels = c("Stem","Leaf"))) %>%arrange(Hybrid, Year, Site, desc(Organ)) %>%group_by(Hybrid, Year, Site) %>%mutate(ymin = cumsum(Value),ymax = cumsum(Value) + sd)# 堆积柱状图（添加标准差，多重比较结果）ggplot(result_plot, aes(x = Hybrid, y = Value, fill = Organ)) +facet_grid(Year ~ Site) +geom_bar(stat = "identity", width = 0.8, aes(group = Organ)) +geom_errorbar(aes(ymin = ymin, ymax = ymax, color = Organ), width = 0.2) +geom_text(aes(y = ymax + 10,x = Hybrid, group = Organ, label = groups)) +scale_y_continuous(expand = c(0,0)) +labs(x = NULL, y = quote(Biomass~(Mg~ha^'-1')),fill = "Organ") +theme_test() +theme(axis.text = element_text(color = 'black'),legend.position = "bottom")
代码中演示数据的result结果
代码中演示数据可视化结果
注意： warning_normal和warning_levene为FALSE表示通过 正态性检验和方差齐性检验， 为TRUE则需要注意，在这种情况下，多重比较依旧采用了指定的方法进行，统计结果需要推敲。另外对于两组变量的比较应该选择T检验，但依旧使用的多重比较（因变量为2组时等同于T检验）。