Morning2

  

  urlfile <-"https://raw.githubusercontent.com/dantalus/intro_workshop/master/plot.csv"
  plotDf <-read.csv(urlfile)
  
# install.packages(c("ggthemes", "ggbeeswarm"))
  library(ggthemes)
  library(ggbeeswarm)
  library(ggplot2)
  library(dplyr)
  
# WTF is a tibble?  
  class(plotDf)
  
  as.tbl(plotDf)  
  
  plotDf <- as.tbl(plotDf)
  
  class(plotDf) <- "data.frame"

  library(readr)    
  plotDf <-read_csv(urlfile)
  
  plotDf <-as_tbl(read.csv(urlfile))
  plotDf <-read.csv(urlfile) %>% as_tbl()
  
# Rename variables
# Base R  
  plotDf$age <- plotDf$demo_age
  plotDf[, -5]
  plotDf[, !grepl("demo_age", names(plotDf))]
  
  names(plotDf)[5] <- "age"
  
# dplyr
  
  plotDf <-read_csv(urlfile)
  
  rename(plotDf, age = demo_age)
  
  
# Select rows ####
# Base R  
  plotDf[1:5, ]
  
  plotDf[plotDf$id > 199, ]
  plotDf[plotDf$id > 199 & plotDf$demo_gender == "Male", ]
  plotDf[grepl(" A", plotDf$arm), ]  # character matching
  
  r <- plotDf$id > 199
  plotDf[r, ]
  
  r <- plotDf$id %in% c(201, 202, 101)
  plotDf[r, ]
  
  subset(plotDf, id > 199)
  
# dplyr  
  
  filter(plotDf, id > 199 & demo_gender == "Male")
  filter(plotDf, id > 199 | demo_gender == "Male")
  filter(plotDf, (id > 199 & demo_gender == "Male") |
                 (id < 199 & demo_gender == "Female"))
  
# Putting rows back together
  
  a <- filter(plotDf, id > 199)
  b <- filter(plotDf, id <= 199)
  
  rbind(a, b)
  
  a <- filter(plotDf, id > 199) %>%
       rename(age = demo_age)
  b <- filter(plotDf, id <= 199)
  
  rbind(a, b) #error
  
  names(b) <- names(a)
  
# select columns ####
  
# Base R  
  
  plotDf$glvef %>% class()
  plotDf[, 1] %>% class()
  plotDf[[1]]
  plotDf[1]
  
  
  plotDf[1, 1]
  plotDf[[1]][1]
  
  
  plotDf[, "glvef"]  
  plotDf[, grepl("demo", names(plotDf))]
  
# dplyr 
  
  select(plotDf, id, glvef)
  select(plotDf, glvef:id)
  
  select(plotDf, starts_with("demo"))
  
  select(plotDf, id, GLVEF = glvef)
  
  plotDf <- select(plotDf, arm, id, time, everything() )
  
# Putting columns back together
  
  a <- select(plotDf, id, glvef)
  b <-select(plotDf, starts_with("demo"))
  
  cbind(a, b) # danger
  
  a <- select(plotDf, id, time, glvef)
  b <- select(plotDf, id, time, starts_with("demo"))
  
  g <- full_join(a, b, by = c("id", "time"))
  
  cbind(a, b) # danger
  
  
# New/modify variables
# Base R  
  plotDf$bmi <- plotDf$demo_wt_kg / (plotDf$demo_ht_cm / 100)^2
  
  plotDf[plotDf$id == 107, ]$demo_ht_cm <- NA
  View(plotDf)
  
  plotDf[plotDf$demo_gender == "Male", ]$demo_age <- NA
  
  plotDf <-read_csv(urlfile)

# dplyr
  
  plotDf <- mutate(plotDf, bmi = demo_wt_kg / (demo_ht_cm / 100)^2)
  
  mutate(plotDf, bmi_rank = percent_rank(bmi)) %>% View()
  
  plotDf <- mutate(plotDf, bmi = demo_wt_kg / (demo_ht_cm / 100)^2) %>%
            mutate(bmi_rank = percent_rank(bmi))
  
# Arrange by rows
  
# Base R
  
  plotDf[order(plotDf$demo_age), ]
  plotDf[order(plotDf$arm, plotDf$id, plotDf$time), ]
  
# dplyr
  
  arrange(plotDf, arm, id, time)
  
# Unique rows
  
  plotDf[!duplicated(plotDf$id), ]
  
  unique(plotDf[c("id")])
  
# dplyr
  
  distinct(plotDf, id, .keep_all = TRUE)
  distinct(plotDf, id) %>% unlist() %>% as.numeric()
  
  
  distinct(plotDf, id, time, .keep_all = TRUE)
  
# Summarizing
  
  summarise(plotDf, 
            n = n(), 
            mean_glvef = mean(glvef, na.rm = TRUE), 
            min = min(glvef, na.rm = TRUE))
  
  group_by(plotDf, arm, demo_gender) %>%
    summarise(n = n(), 
              mean_glvef = mean(glvef, na.rm = TRUE))
  
# Group, summarise, join  
  group_by(plotDf, arm) %>%
    summarise(n = n(), 
              mean_glvef = mean(glvef, na.rm = TRUE)) %>%
    full_join(plotDf, by = "arm") %>% View()
  
# Group, mutate, join
  group_by(plotDf, demo_gender) %>%
    mutate(bmi_rank = percent_rank(demo_ht_cm)) %>%
    full_join(plotDf, by = "demo_gender")  %>% View()
  
  
# Back to our plot ####
  
  plotDf <-read.csv(urlfile)
  
  plotDf <- mutate(plotDf, arm = factor(arm, labels = c("Placebo", "Low Dose", "High Dose"))) %>%
    mutate(time = factor(time, levels = c("Baseline", "8 weeks"))) 
  
  
  ggplot(plotDf, aes(y = glvef, shape = arm)) +
    geom_violin(aes(x = as.numeric(time), group = time),
                color = "grey90", fill = "grey90", width = .5) +
    geom_line(aes(group = id,
                  x = as.numeric(time)),
              alpha = 0.7, linetype = "dashed") +
    facet_wrap(~arm) +
    theme_base() +
    scale_x_continuous(breaks = c(1, 2),
                       labels = c("Baseline", "8 weeks")) +
    geom_point(aes(group = id, x = as.numeric(time))) +
    geom_smooth(aes(x = as.numeric(time)),
                method = "lm", se = FALSE, size = 2, color = "black") +
    geom_pointrange(data = filter(plotDf, arm == "Placebo" & time == "Baseline"),
                    aes(x = 0.5,
                        y =    mean(glvef, na.rm = TRUE),
                        ymax = mean(glvef, na.rm = TRUE) +
                          sd(glvef, na.rm = TRUE),
                        ymin = mean(glvef, na.rm = TRUE) -
                          sd(glvef, na.rm = TRUE))) +
    geom_pointrange(data = filter(plotDf, arm == "Placebo" & time == "8 weeks"),
                    aes(x = 2.5,
                        y =    mean(glvef, na.rm = TRUE),
                        ymax = mean(glvef, na.rm = TRUE) +
                          sd(glvef, na.rm = TRUE),
                        ymin = mean(glvef, na.rm = TRUE) -
                          sd(glvef, na.rm = TRUE))) +
    geom_pointrange(data = filter(plotDf, arm == "Low Dose" & time == "Baseline"),
                    aes(x = 0.5,
                        y =    mean(glvef, na.rm = TRUE),
                        ymax = mean(glvef, na.rm = TRUE) +
                          sd(glvef, na.rm = TRUE),
                        ymin = mean(glvef, na.rm = TRUE) -
                          sd(glvef, na.rm = TRUE))) +
    geom_pointrange(data = filter(plotDf, arm == "Low Dose" & time == "8 weeks"),
                    aes(x = 2.5,
                        y =    mean(glvef, na.rm = TRUE),
                        ymax = mean(glvef, na.rm = TRUE) +
                          sd(glvef, na.rm = TRUE),
                        ymin = mean(glvef, na.rm = TRUE) -
                          sd(glvef, na.rm = TRUE))) +
    geom_pointrange(data = filter(plotDf, arm == "High Dose" & time == "Baseline"),
                    aes(x = 0.5,
                        y =    mean(glvef, na.rm = TRUE),
                        ymax = mean(glvef, na.rm = TRUE) +
                          sd(glvef, na.rm = TRUE),
                        ymin = mean(glvef, na.rm = TRUE) -
                          sd(glvef, na.rm = TRUE))) +
    geom_pointrange(data = filter(plotDf, arm == "High Dose" & time == "8 weeks"),
                    aes(x = 2.5,
                        y =    mean(glvef, na.rm = TRUE),
                        ymax = mean(glvef, na.rm = TRUE) +
                          sd(glvef, na.rm = TRUE),
                        ymin = mean(glvef, na.rm = TRUE) -
                          sd(glvef, na.rm = TRUE))) +
    # scale_color_brewer(guide = FALSE, palette = "Set1") +
    # scale_fill_manual(guide  = FALSE, palette = "Set1") +
    scale_shape(guide = FALSE) +
    xlab("")+
    ylab("GLVEF (%)") +
    ylim(0, 75) +
    theme(panel.spacing = unit(2, "lines"),
          panel.border = element_rect(color = "white"),
          axis.text.y = element_text(size = 16),
          strip.text = element_text(size = 16))