Example code used for the examples at http://emilkirkegaard.dk/en/?p=5926

Missing data example R code

library(pacman)
p_load(kirkegaard, magrittr, ggplot2)

# example dataset ---------------------------------------------------------

set.seed(1)
iris_miss = df_addNA(iris)

# amount ------------------------------------------------------------------

miss_amount(iris_miss)

miss_by_case(iris_miss)

miss_by_case(iris_miss) %>% table()

miss_by_var(iris_miss)

miss_plot(iris_miss)
ggsave("miss_plot_case.png")
miss_plot(iris_miss, case = F)
ggsave("miss_plot_var.png")


# pattern -----------------------------------------------------------------
iris_miss[1:10, ]
miss_matrix(iris_miss[1:10, ])

iris_miss[miss_by_case(iris_miss) >= 1, ] %>% head(10)


miss_pattern(iris_miss)
miss_pattern(iris_miss) %>% table()

miss_complexity(iris_miss)



# dependencies ------------------------------------------------------------

miss_analyze(iris_miss)
miss_analyze(iris_miss) %>% MAT_get_half() %>% abs() %>% psych::describe()

set.seed(1)
rnorm(5e3) %>% matrix(nrow = 1000) %>% as.data.frame() %>% df_addNA() %>% miss_analyze()
set.seed(1)
rnorm(5e3) %>% matrix(nrow = 1000) %>% as.data.frame() %>% df_addNA() %>% miss_analyze()  %>% MAT_get_half() %>% abs() %>% psych::describe()