Code for workshop on R for archaeologists at the National Taiwan University as part of my BITSS series

BITSS-NTU-workshop.R

# load the libraries
library(tidyverse)
library(readxl)

# load the data
pottery <- 
  read_excel("data/HTSKY-RIM.xlsx",
             col_types = "text")

# inspect the data
View(pottery)
str(pottery)
names(pottery)

# clean the data
pottery_clean <- 
  pottery %>% 
  mutate_at(c("RimD", "LipT", "NeckH", "Weight", "Thickness"),
            as.numeric)

# analyse a little
pottery_clean %>% 
  count(Area, sort = TRUE)

pottery_clean %>% 
  count(Sarea, sort = TRUE)


# plot some data
ggplot(pottery_clean,
       aes(Weight)) +
  geom_histogram() +
  scale_x_log10()

ggplot(pottery_clean,
       aes(NeckH,
           LipT)) +
  geom_point() +
  theme_minimal() 

ggplot(pottery_clean,
       aes(Slayer,
           LipT)) +
  geom_boxplot() +
  geom_jitter() +
  theme_minimal() 


# plot many variables together
pottery_clean %>% 
  select(Slayer, RimD:Weight) %>% 
  gather(variable, value, -Slayer) %>% 
  filter(!is.na(Slayer)) %>% 
  ggplot(aes(Slayer,
             value)) +
  geom_boxplot() +
  geom_jitter(alpha = 0.3) +
  facet_wrap( ~ variable, scales = "free_y") +
  theme_minimal()

# save the plot
ggsave("pottery_metrics_by_layer.png")

# statistical tests

# ANOVA
aov_lipT_by_layer <- 
aov(data = pottery_clean,
    LipT ~ Slayer) 

summary(aov_lipT_by_layer)
TukeyHSD(aov_lipT_by_layer)

# chi-square
# chi-sq
pottery_clean_table <- 
  pottery_clean %>% 
  count(Rim, Area) %>% 
  spread(Rim, n) %>% 
  select(-Area) %>% 
  filter_all(any_vars(. > 5)) %>% 
  select_if( ~!any(is.na(.))) %>% 
  select(-`<NA>`)

chisq.test(pottery_clean_table)