failed when I was trying to use defuse-inject pattern in dplyr group_map function - r

Question

Following is an example of what I am trying to achieve.
library(dplyr)
tbl.data <- tidyquant::tq_get(c("GS", "C", "BAC"))
to.xts <- function(group, group_key, date_col, price_col){
a <- group %>% dplyr::pull({{ price_col }})
b <- group %>% dplyr::pull({{ date_col }})
x <- xts::xts(a, order.by=b)
colnames(x) <- key$symbol
x
}
make.xts <- function(data, date_col, price_col){
data %>%
group_by(symbol) %>%
group_map(~to.xts(.x, .y, date_col, price_col))
}
# Failed example one:
tbl.data %>% group_by(symbol) %>% group_map(to.xts, date, close)
# Failed example two:
make.xts(tbl.data, date, close)
# Error in `dplyr::pull()`:
# ! Can't extract column with `!!enquo(var)`.
# ✖ `!!enquo(var)` must be numeric or character, not a function.
# Run `rlang::last_error()` to see where the error occurred.
# However, If I single out a group myself and apply `to.xts` to that group it'll work. The only thing changed, which I doubt that it would have effect on the function itself, is that the `group_key` is now a string (it was a data-variable in the context of `group_map`'s `.f`)
gs.grp <- tbl.data %>% dplyr::filter(symbol=="GS")
gs.grp %>% to.xts("GS", date, col)
# A simply pull operation would also work.
gs.grp %>% dplyr::pull(close)
I don't quite understand what has changed internally; why this is the case and what's not correct here?
Given the error message that it seems dplyr::pull is doing
defuse (enquo) and inject (!!) itself internally, therefore I shall not use embracing operator; however, without it it didn't work either and caused the same error.

Answer 1

I haven't used group_map function much , here is an alternative version that you can try -
library(dplyr)
library(purrr)
tbl.data <- tidyquant::tq_get(c("GS", "C", "BAC"))
to.xts <- function(group, symbol, date, price){
a <- group %>% dplyr::pull({{ price }})
b <- group %>% dplyr::pull({{ date }})
x <- xts::xts(a, order.by=b)
colnames(x) <- symbol
x
}
tbl.data %>% split(.$symbol) %>% imap(~to.xts(.x, .y, date, close))
If you want them in one xts object as separate column.
tbl.data %>%
split(.$symbol) %>%
imap(~to.xts(.x, .y, date, close)) %>%
{do.call(merge, .)}
# BAC C GS
#2013-01-02 12.03 41.25 131.66
#2013-01-03 11.96 41.39 130.94
#2013-01-04 12.11 42.43 134.51
#2013-01-07 12.09 42.47 134.26
#2013-01-08 11.98 42.46 133.05
#2013-01-09 11.43 42.04 134.32

Answer 2

Using group_map using a formula to apply to each group:
library(dplyr)
#library(xts)
tbl.data %>%
group_by(symbol) %>%
group_map(~ to.xts(.x, .y, date, close))
to.xts <- function(group, group_key, date, price) {
a <- group %>% dplyr::pull({{ price }})
b <- group %>% dplyr::pull({{ date }})
x <- xts::xts(a, order.by = b)
colnames(x) <- group_key
x
}
Very similar notation to purr that it is based on.
In the formula, you can use
. or .x to refer to the subset of rows of .tbl for the given group
.y to refer to the key, a one row tibble with one column per grouping variable that identifies the group
(See the documentation)
Alternatively, we could also use pivoting to avoid your function, and put it inside one xts-object.
library(dplyr)
library(tidyr)
#library(xts)
tbl.data %>%
pivot_wider(date, names_from = symbol, values_from = low) %>%
xts::xts(order.by = .$date) %>%
.[,-1]
Output:
GS C BAC
2013-01-02 "130" "40.7" "11.9"
2013-01-03 "130" "41.0" "11.9"
2013-01-04 "130" "41.6" "11.9"
2013-01-07 "133" "42.0" "12.0"
2013-01-08 "133" "42.0" "11.9"
2013-01-09 "133" "41.8" "11.3"
2013-01-10 "134" "42.0" "11.5"

Answer 3

This doesn't answer your question, but I wanted to point out that this is much less complicated if you avoid the tidyverse patterns completely.
symbols <- c("GS", "C", "BAC")
# Environment to hold data
my_data <- new.env()
# Tell getSymbols() to load the data into 'my_data'
getSymbols(symbols, env = my_data)
# Combine all the close prices into one xts object
price_data <- Reduce(merge, lapply(my_data, Cl))
# Remove ".Close" from column names
colnames(price_data) <- sub(".Close", "", colnames(price_data), fixed = TRUE)