# solution WS 3 Graphs

# importing data
library(readxl)
data = read_excel("path")
#  categorical variables
mgus2$sex <- factor(mgus2$sex,
                   levels = c('F', 'M'),
                   labels = c('female', 'male'))
mgus2$pstat <- factor(mgus2$pstat,
                     levels = c('0', '1'),
                     labels = c('no', 'yes'))
mgus2$death <- factor(mgus2$death,
                     levels = c('0', '1'),
                     labels = c('no', 'yes'))

#  continuous variables 
mgus2$hgb <- as.numeric(mgus2$hgb)
mgus2$creat <- as.numeric(mgus2$creat)
mgus2$mspike <- as.numeric(mgus2$mspike)

# categorical variable "fu_cat"
data$fu_cat[data$futime < 60] <- 1
data$fu_cat[(data$futime >= 60) & (data$futime < 120)] <- 2
data$fu_cat[(data$futime >= 120) & (data$futime < 180)] <- 3
data$fu_cat[data$futime >= 180] <- 4
# declaring new variable as categorial
data$fu_cat <- factor(data$fu_cat, 
                      levels = c('1', '2', '3', '4'),
                      labels = c('< 5 yrs', '5 - 10 yrs', '10 - 15 yrs', '> 15 yrs'))


# ex. 1
# a. pie chart for fu_cat
library(ggplot2)
#  create data frame with categorial variable
df <- data.frame(table(data$fu_cat))
names(df)[names(df) == "Var1"] <- "futime"
names(df)[names(df) == "Freq"] <- "frequency"
df$percent <- round(df$frequency/sum(df$frequency) * 100, 2) # add column with percentages 
#  create piechart with ggplot
pie <- ggplot(df, aes(x = "", y = frequency, fill = futime)) + 
  coord_polar("y", start = 0) +
  geom_bar(width = 1, stat = "identity") +
  geom_text(aes(label = paste(df$percent, "%")),
            position = position_stack(vjust = 0.5)) +
  ggtitle("follow-up time (death)") +
  theme(plot.title = element_text(hjust = 0.5)) 
  
#  print piechart
pie

# b. barplot for creat, separating age_qt and sex
#  categorical variable from age
data$age_qt[data$age < 63] <- 1
data$age_qt[(data$age >= 63) & (data$age < 72)] <- 2
data$age_qt[(data$age >= 72) & (data$age < 79)] <- 3
data$age_qt[data$age >= 79] <- 4
#  declare as categorical and change labels
data$age_qt <- factor(data$age_qt, 
                      levels = c('1', '2', '3', '4'),
                      labels = c('< 63 yrs', '63 - 71 yrs', '72 - 78 yrs', '> 78 yrs'))

#  create barplot
plot <- ggplot(data = data, 
               aes(x = age_qt, y = creat, fill = sex)) + 
  geom_bar(stat = "identity", 
           position = position_dodge()) +
  ggtitle("creatinine for age groups, split by sex") + 
  theme(plot.title = element_text(hjust = 0.5))
#  print plot
plot

# ex. 2
# a. advanced histogram for age 
ggplot(data = data, aes(x = age)) + 
  geom_histogram(aes(y = ..density..), 
                 bins = 50) + # number of bars
  geom_density()

# b. boxplots for hgb
par(mfrow = c(1, 3))
female <- subset(mgus2, mgus2$sex == "female")
male <- subset(mgus2, mgus2$sex == "male")
boxplot(mgus2$hgb,
        # range = max(data$hgb, na.rm = TRUE), # determines how far the whisker extends. if set to 0 extends to both extremes
        main = "overall hgb levels",
        ylab = "hgb (g / dl)")
boxplot(female$hgb,
        # range = max(female$hgb, na.rm = TRUE),
        main = "hgb levels females",
        ylab = "hgb (g / dl)")
boxplot(male$hgb,
        # range = max(male$hgb, na.rm = TRUE),
        main = "hgb levels males",
        ylab = "hgb (g / dl)")
mtext("boxplots for haemoglobin, split by sex", side = 3, outer = TRUE, cex = 0.9)

#  other option for nicer plot
ggplot(data = mgus2, aes(x = sex, y = hgb)) + # x: factor, y: variable
  geom_boxplot() + # produces a boxplot with data from above
  geom_jitter(colour = "black", size = 1, shape = 1) + # adds data points
  stat_summary(fun.y = mean, # adds mean to plot
               geom = "point", 
               shape = 17, size = 4, col = "red") # as red triangle (shape 17)

# c. to change whiskers of histograms, remove '#' before range in boxplots

# d. qq-plot creatinine
par(mfrow = c(1, 1))
qqnorm(data$creat, pch = 1, frame = FALSE)
qqline(data$creat, col = "steelblue", lwd = 2)
#  data is not normally distributed since the model does not fit the data


# ex. 3
# a. import data
rats_weight = read_excel("E:/Malawi 2024/Summer School Malawi/Data/ratsweight.xlsx")

# b. convert to factor
rats_weight$Diet <- factor(rats_weight$Diet)

# c. spaghetti plot
spaghetti <- ggplot(data = rats_weight, aes(x = Time, y = weight, group = Rat))
spaghetti + 
  geom_line() + 
  facet_grid(. ~ Diet) # add this parameter for d. separated plots