Partial grouping inside a dataframe in R-CodePudding

For statistical analysis purpose, I would like to regroup some rows inside a data frame based on their values.

What I have:

number	latitude
30	57
12	59
01	68
12	66
101	55
47	61
05	60
288	67

The desired output would be, for example, to regroup every latitude above 66 (66 67 68) in a single category 66 and the desired output would be like this:

number	latitude	new
30	57	57
12	59	59
01	68	66
12	66	66
101	55	55
47	61	61
05	60	60
288	67	66

I do not want to use an if loop because I feel that it is not really R friendly. I would also like to keep the initial column, that way I can try different combinations later on.

Thank you very much.

CodePudding user response：

Option mutate and ifelse:

library(dplyr)
df %>%
  mutate(new = ifelse(latitude >= 66, "66 ", latitude))

Output:

  number latitude new
1     30       57  57
2     12       59  59
3     01       68 66 
4     12       66 66 
5    101       55  55
6     47       61  61
7     05       60  60
8    288       67 66

Data

df <- data.frame(number = c("30","12","01","12","101","47","05","288"),
                 latitude = c(57,59,68,66,55,61,60,67))

CodePudding user response：

library(tidyverse)

tribble(~"number",  ~"latitude",
        30, 57,
        12, 59,
        01, 68,
        12, 66,
        101,55,
        47, 61,
        05, 60,
        288,67) %>% 
  dplyr::mutate(
    new = if_else(latitude > 66,
                  "66 ",
                  as.character(latitude)))

CodePudding user response：

We can use

df1$new <- df1$latitude
df1$new[df1$latitude >=66] <- "66 "

or with ifelse

df1$new <- with(df1, ifelse(latitude >=66, "66 ", latitude))

-output

> df1
  number latitude new
1     30       57  57
2     12       59  59
3      1       68 66 
4     12       66 66 
5    101       55  55
6     47       61  61
7      5       60  60
8    288       67 66

Also, as @Mael commented about the type of 'new' column, if we want to preserve the type, can also use pmin

library(dplyr)
df1 %>%
    mutate(new = pmin(latitude, 66))
   number latitude new
1     30       57  57
2     12       59  59
3      1       68  66
4     12       66  66
5    101       55  55
6     47       61  61
7      5       60  60
8    288       67  66