glueformula: simply build regression formulas from vectors with variable names

This article is originally published at http://skranz.github.io/

The small new package glueformula with a single function gf facilitates constructing regression formulas from vectors with variable names. The syntax is similar to glue strings. Here is an example:

# Example: build a formula
# for ivreg with gf
library(glueformula)

# Contol variables and instruments
contr = c("x1","x2","x3","log(x4)")
instr = c("z1","z2",contr)

# Create formula for ivreg
gf(q ~ p + {contr} | {instr})

## q ~ p + x1 + x2 + x3 + log(x4) | z1 + z2 + x1 + x2 + x3 + log(x4)

There is no big benefit if one wants to estimate a single regression, but in econometrics one often specifies a lot of similar regressions (for robustness checks not p-hacking!) that share a large set of common control variables. Here glueformula can be handy.

You can install it from Github as explained here: https://github.com/skranz/glueformula

There was also a discussion here whether a similar feature should be included into the glue package itself, but it looks as if that is not going to happen.

Thanks for visiting r-craft.org
This article is originally published at http://skranz.github.io/
Please visit source website for post related comments.