Extract first component of a string list: dplyr vs. purrr

Goal: Compare two styles of functional programming on parsing string.
Input: The cabin column in Titanic data (Kaggle)
Output: the number extracted from the first components of the space delimited string


Code:
#################################################
library(microbenchmark)
library(tidyverse)
library(stringr)

pc <- microbenchmark(
    purrrV  = {  
        trainDf0 <- trainDfRaw
        trainDf0$Cabin %>%
            as.character() %>%
            map(str_split, " ") %>%
            flatten() %>%
            map_chr(1) %>%
            map_chr(str_replace_all, '[[:alpha:]]','') %>%
            map(as.integer) %>%
            map(coalesce, as.integer(0)) -> trainDf0$Cabin_Number
    },
    dplyV = {
        trainDf1 <- trainDfRaw
        trainDf1 %>% mutate(Cabin_Number = as.character(Cabin)) %>%
            separate(Cabin_Number, into = c("Cabin_Number"), sep = " ",  extra = "drop", remove = TRUE) %>%
            mutate(Cabin_Number = coalesce(as.integer(str_replace(Cabin_Number, "[[:alpha:]]", "")), as.integer(0))) ->
            trainDf1
    }
)

Result
Unit: milliseconds
   expr        min         lq       mean     median        uq        max neval
 purrrV 1363.32589 1405.10168 1607.33881 1429.10920 1561.6362 3453.50236   100
  dplyV   15.69765   16.69651   19.83311   17.46755   19.8518   79.28219   100


dply is much faster than purrr