Data Manipulation Tools

Overview

The R programming language has extensive functionality and the vast ecosystem of community-developed packages extends those capabilities well beyond the base language. Nonetheless, there is always room for additional helpful functions to be added to any R programmer’s toolbelt and the sections below discuss some of those available within thekidsbiostats.

Example Usage

Rounding Functions

The following two functions (round_vec and round_df) are examples where the sole purpose is to keep code concise and to the point. Often when we present figures — whether that be through plots, tables or in written form — we wish to preserve trailing zeroes when rounding numbers to a specific x decimal places.

There are two functions that can help with this:

• round_vec – preserves trailing zeroes for vectors.
• round_df – consistently rounds every numeric column in a data.frame or tibble.

Below we demonstrate the difference between the base R round() and round_vec() when rounding numerical values:

original <-  c(1.8003, 1.9998, 2.5812)

data.frame(original = original) %>%
  mutate(round     = as.character(round(original, 2)),
         round_vec = round_vec(original, 2)) %>%
  thekids_table()

original	round	round_vec
1.8003	1.8	1.80
1.9998	2	2.00
2.5812	2.58	2.58

To illustrate round_df:

data.frame(var1 = rnorm(n = 5, mean = 10, sd = 2),
           var2 = rexp(n = 5, rate = 0.25),
           var3 = rweibull(n = 5, shape = 4, scale = 7))  %>%
  round_df(digits = 2) %>%
  thekids_table()

var1	var2	var3
7.20	0.09	9.02
10.51	3.90	6.90
5.13	5.21	2.78
9.99	7.81	7.38
11.24	5.23	5.52

Data Manipulation Functions

> fct_case_when

Factors are a useful data format for manipulating any categorical data because they preserve the ordinal nature inherent to those variables.

Below is an example of working with ordinal data.

x <- 1:50
case_when(x %% 12 == 0 ~ "Very Likely",   # Multiple of 12 (most certain)
          x %% 6  == 0 ~ "Likely",        # Multiple of 6
          x %% 3  == 0 ~ "Neutral",       # Multiple of 3
          x %% 2  == 0 ~ "Unlikely",      # Multiple of 2
          TRUE         ~ "Very Unlikely"  # Default category) %>% 
          ) %>% 
  as.factor %>%
  levels()
#> [1] "Likely"        "Neutral"       "Unlikely"      "Very Likely"  
#> [5] "Very Unlikely"

Note, however, the strange (alphabetical) ordering of the levels. If we would like to set a more logical ordering of these factors, we would also have to use factor:

x <- 1:50
case_when(x %% 12 == 0 ~ "Very Likely",   # Multiple of 12 (most certain)
          x %% 6  == 0 ~ "Likely",        # Multiple of 6
          x %% 3  == 0 ~ "Neutral",       # Multiple of 3
          x %% 2  == 0 ~ "Unlikely",      # Multiple of 2
          TRUE         ~ "Very Unlikely"  # Default category) %>% 
          ) %>% 
  factor(levels = c("Very Unlikely", "Unlikely", "Neutral", "Likely", "Very Likely")) %>%
  as.factor %>%
  levels()
#> [1] "Very Unlikely" "Unlikely"      "Neutral"       "Likely"       
#> [5] "Very Likely"

In comparison, fct_case_when() orders the factor levels simply based on the order of their appearance in the argument, to return an identical result to the above:

x <- 1:50
fct_case_when(x %% 12 == 0 ~ "Very Likely",   # Multiple of 12 (most certain)
              x %% 6  == 0 ~ "Likely",        # Multiple of 6
              x %% 3  == 0 ~ "Neutral",       # Multiple of 3
              x %% 2  == 0 ~ "Unlikely",      # Multiple of 2
              TRUE         ~ "Very Unlikely"  # Default category) %>% 
              ) %>% 
  as.factor %>%
  levels()
#> [1] "Very Likely"   "Likely"        "Neutral"       "Unlikely"     
#> [5] "Very Unlikely"