convert_cols: Convert dataframe column types.
In curtisalexander/CRAmisc: Curtis Miscellaneous

Description Usage Arguments Details Value Examples

Given a configuration dataframe with column names and the required types, convert the column data types in a separate dataframe.

1	convert_cols(df, types_df)

`df`	A dataframe that will have its column data types converted.
`types_df`	A dataframe containing the column to convert and the data type to convert to. The expectation is that `types_df` has columns named `col_name` and `col_type`. If `types_df` has columns named differently, this is easy to convert using rename.

Column metadata may be stored separately from the code needed to perform the conversion. As an example, a dataframe with 2,000 columns could have its metadata stored in a separate csv or xlsx file. The metadata file would be read in as a dataframe and convert_cols would update the data types en masse.

convert_cols is trading the need to maintain a separate metadata file for concise code that performs the data type conversion. The alternative is to have verbose code that details each data type conversion to take place.

convert_cols is a closure that contains two parts.

An inner function, convert_, that converts a set of columns that are to be converted to the same type using dmap_at.
The enclosing environment that contains the dataframe that will be updated and contains a call to pmap to drive iteration.

Finally, the function is missing cases to convert to the following types. The types includes should cover the most simple use cases.

list
factor
raw
complex

A copy of the original dataframe, df, with the column types converted.

## dataframe to convert
id <- c(1L, 2L, 3L)
date_var <- c("2016-01-02", "2015-12-24", "2016-05-05")
datetime_var <- c("2016-01-02T11:11:04",
                  "2015-12-24T08:11:41",
                  "2016-05-05T05:05:49")
double_var1 <- c("46.41", "118.11", "84.68")
double_var2 <- c("68.48", "-248.99", "194")
numeric_var <- c("78.61", "593.1", "123")
logical_var <- c("true", "true", "false")
character_var <- c(49, 88, 104)
integer_var <- c("77", "84", "4949")

test_df <- data.frame(id,
                      date_var,
                      datetime_var,
                      double_var1,
                      double_var2,
                      numeric_var,
                      logical_var,
                      character_var,
                      integer_var)

## print the structure
str(test_df)

## create a dataframe with columns named col_name and col_type (if you name
##   them something else you can always use dplyr::rename) with the column
##   name and the type you would like it converted to
types_df <- tibble::tribble(
  ~col_name, ~col_type,
  "date_var", "date",
  "datetime_var", "datetime",
  "double_var1", "double",
  "double_var2", "double",
  "numeric_var", "numeric",
  "logical_var", "logical",
  "character_var", "character",
  "integer_var", "integer"
)

## convert the columns
test_df_converted <- convert_cols(test_df, types_df)