使用str_extract从stringr中提取数字。

library(stringr)

cbind(df, col3 = as.numeric(str_extract(df$col2, "[\\d+.]+")))
   col1            col2  col3
1    t1             300 300.0
2    t2           >200m 200.0
3    t3      wef3.2 wef   3.2
4    t4  result 50 mg/g  50.0
5    t5            <NA>    NA
6    t6        Not data    NA
7    t7         pending    NA
8    t8            <NA>    NA
9    t9        positive    NA
10  t10 data >20 mile/h  20.0

使用gsub，删除除数字以外的所有内容。

cbind(df, col3 = as.numeric(
  gsub("([.-])|[[:alpha:][:punct:] ]", "\\1", df$col2)))
   col1            col2  col3
1    t1             300 300.0
2    t2           >200m 200.0
3    t3      wef3.2 wef   3.2
4    t4  result 50 mg/g  50.0
5    t5            <NA>    NA
6    t6        Not data    NA
7    t7         pending    NA
8    t8            <NA>    NA
9    t9        positive    NA
10  t10 data >20 mile/h  20.0

或者使用\\D（非数字）代替[:alpha:]和[:punct:]（感谢@thelatemail和@onyambu）!

数据

df <- structure(list(col1 = c("t1", "t2", "t3", "t4", "t5", "t6", "t7", 
"t8", "t9", "t10"), col2 = c("300", ">200m", "wef3.2 wef", "result 50 mg/g", 
NA, "Not data", "pending", NA, "positive", "data >20 mile/h")), 
row.names = c(NA, -10L), class = "data.frame")

一个可能的选项是使用readr package中的parse_number()，例如

library(readr)
col1 <- c('t1', 't2', 't3', 't4', 't5', 't6', 't7', 't8', 't9', 't10')
col2 <- c(300, '>200m', NA, 'result 50 mg/g', NA, 'Not data', 'pending', NA, 'positive', 'data >20 mile/h')
df <- data.frame(col1, col2)

df$col3 <- parse_number(df$col2)
#> Warning: 3 parsing failures.
#> row col expected   actual
#>   6  -- a number Not data
#>   7  -- a number pending 
#>   9  -- a number positive
df
#>    col1            col2 col3
#> 1    t1             300  300
#> 2    t2           >200m  200
#> 3    t3            <NA>   NA
#> 4    t4  result 50 mg/g   50
#> 5    t5            <NA>   NA
#> 6    t6        Not data   NA
#> 7    t7         pending   NA
#> 8    t8            <NA>   NA
#> 9    t9        positive   NA
#> 10  t10 data >20 mile/h   20

创建于2023年2月7日，使用reprex v2.0.2