loads: Loading only given variables of a data.frame from binary file

Description

loads does what the name suggests: it loads data from a special binary file format (RDatas) made up by the saves function. This special, uncompressed tar archive inlcudes several separate RData files (saved by save function) as being columns/variables of a data.frame.

Usage

loads(file = NULL, variables = NULL, to.data.frame = FALSE, ultra.fast = FALSE)

Arguments

file

character string: the (RDatas) filename from which to load the variables. If using

ultra.fast =
  TRUE

option, specify the directory holding the uncompressed R objects (saved via

saves(...,
  ultra.fast = TRUE)

variables

A character vector containing the variable names to load

to.data.frame

boolean: the default behavior of loads is to concatenate the variables to a list. This could be overriden with TRUE argument specified at to.data.frame parameter, which will return a dataframe instead of list. Only do this if all your variables have the same number of cases!

ultra.fast

boolean: if TRUE, ultra fast (...) processing is done without any check to parameters or file existence/permissions. Be sure if using this setting as no debugging is done! Only recommended for servers dealing with lot of R objects' saves and loads in a monitored environment. Also, for performance gain, it is advised not to convert the list to data frame (to.data.frame = FALSE).

Value

Loaded data.frame

Details

The purpose of this function is to be able only a few variables of a data.frame really fast. It is done by reading and writing datas in binary format without any transformations, and combining the speed of only reading the needed part of an archive.

Some minor experiments shows a huge performance gain against using SQLite/MySQL backends or loading whole binary data, but be conscious always choosing the aprropriate method to write and read data.

The author of this package would like to emphasize: this package could be useful only in few cases!

Examples

Run this code

## Not run: 
# # Loading the 'v1' and 'v5' variables of the demo dataset.
# data(evs.2000.hun)
# saves(evs.2000.hun)
# evs.filtered.list <- loads("evs.2000.hun.RDatas", c('v1', 'v5'))
# evs.filtered.df <- loads("evs.2000.hun.RDatas", c('v1', 'v5'), to.data.frame=TRUE)
# ## End(Not run)