hdm: Interfaces for hdm package for data science pipelines.

In intubate: Interface to Popular R Functions for Data Science Pipelines

Description

Interfaces to hdm functions that can be used in a pipeline implemented by magrittr.

Usage

ntbt_rlasso(data, ...)
ntbt_rlassoATE(data, ...)
ntbt_rlassoATET(data, ...)
ntbt_rlassoLATE(data, ...)
ntbt_rlassoLATET(data, ...)
ntbt_rlassoEffects(data, ...)
ntbt_rlassoIV(data, ...)
ntbt_rlassologit(data, ...)
# ntbt_tsls(data, ...)           ## Already defined in sem

Arguments

data	data frame, tibble, list, ...
...	Other arguments passed to the corresponding interfaced function.

Details

Interfaces call their corresponding interfaced function.

Value

Object returned by interfaced function.

Author(s)

Roberto Bertolusso

Examples

## Not run: 
library(intubate)
library(magrittr)
library(hdm)


## ntbt_rlasso: rlasso: Function for Lasso estimation under homoscedastic
##                      and heteroscedastic non-Gaussian disturbances
set.seed(1)
n <- 100 #sample size
p <- 100 # number of variables
s <- 3 # nubmer of variables with non-zero coefficients
X <- Xnames <- matrix(rnorm(n*p), ncol=p)
colnames(Xnames) <- paste("V", 1:p, sep="")
beta <- c(rep(5,s), rep(0,p-s))
Y <- X %*% beta + rnorm(n)

dta <- list(Y = Y, Xnames = Xnames)
rm(Y, Xnames)

## Original function to interface
rlasso(Y ~ Xnames, data = dta)

## The interface puts data as first parameter
ntbt_rlasso(dta, Y ~ Xnames)

## so it can be used easily in a pipeline.
dta %>%
  ntbt_rlasso(Y ~ Xnames)


## Functions for estimation of treatment effects
## ntbt_rlassoATE, ntbt_rlassoATET, ntbt_rlassoLATE, ntbt_rlassoLATET
## do not have examples of use in help.

## Original function to interface
## The interface puts data as first parameter
## so it can be used easily in a pipeline.



## ntbt_rlassoEffects: rigorous Lasso for Linear Models: Inference
set.seed(1)
n <- 100 #sample size
p <- 100 # number of variables
s <- 3 # nubmer of non-zero variables
X <- matrix(rnorm(n*p), ncol=p)
colnames(X) <- paste("X", 1:p, sep="")
beta <- c(rep(3,s), rep(0,p-s))
y <- 1 + X %*% beta + rnorm(n)
data <- data.frame(cbind(y,X))
colnames(data)[1] <- "y"
fm <- paste("y ~", paste(colnames(X), collapse="+"))
fm <- as.formula(fm)                 
rm(y, X)

## Original function to interface
rlassoEffects(fm, I = ~ X1 + X2 + X3 + X50, data=data)

## The interface puts data as first parameter
ntbt_rlassoEffects(data, fm, I = ~ X1 + X2 + X3 + X50)

## so it can be used easily in a pipeline.
data %>%
  ntbt_rlassoEffects(fm, I = ~ X1 + X2 + X3 + X50)



## ntbt_rlassoIV: Post-Selection and Post-Regularization Inference
##                in Linear Models with Many Controls and Instruments
## The example uses non-formula variant. Please see note below about
## possible problem.
data(EminentDomain)
dta <- list(z = EminentDomain$logGDP$z, # instruments
            x = EminentDomain$logGDP$x, # exogenous variables
            y = EminentDomain$logGDP$y, # outcome varialbe
            d = EminentDomain$logGDP$d) # treatment / endogenous variable
str(dta)
## Original function to interface
attach(dta)
rlassoIV(x=x, d=d, y=y, z=z, select.X=FALSE, select.Z=TRUE) 
detach()

## The interface puts data as first parameter
## NOTE: BE CAREFUL (in general in situations as follow)
## The parameter name "d" in this function can result in a nightmare
## (it got me scratching my head for quite a bit).
## In fact, if you call with parameter names (but not naming data)
## call the following version (commented out)
# ntbt_rlassoIV(dta, x=x, d=d, y=y, z=z, select.X=FALSE, select.Z=TRUE)
## there will be an error, as R will expand "d" to "data", and use
## its info (d) instead of dta.
## Right now I am not sure how to manage this situation and avoid
## that unwanted expansion. I will get back to this later.
## To avoid problems you should specify "data" as below
ntbt_rlassoIV(data=dta, x=x, d=d, y=y, z=z, select.X=FALSE, select.Z=TRUE)
## but of course this beats the purpose (we do not want to name "data"),
## and you *cannot* do it in the pipeline version (as you do not include data
## in your call).
## SOLUTION. In cases of unfortunate parameter names: "d", "da", "dat",
## you need to make sure that that parameter is sent by position AND unnamed
ntbt_rlassoIV(dta, x=x, d, y=y, z=z, select.X=FALSE, select.Z=TRUE)
## In general, required data is sent unnamed and by position, like
ntbt_rlassoIV(dta, x, d, y, z, select.X = FALSE, select.Z = TRUE)
## and this would have been what I would have done if I would
## (have had the ability to) produce this example.
## But this is how the example was provided, giving an opportunity to
## uncover this potentially unpleasant situation.

## so it can be used easily in a pipeline.
dta %>%
  ntbt_rlassoIV(x, d, y, z, select.X = FALSE, select.Z = TRUE)


## ntbt_rlassologit: Function for logistic Lasso estimation
library(hdm)
## DGP
set.seed(2)
n <- 250
p <- 100
px <- 10
X <- matrix(rnorm(n*p), ncol=p)
beta <- c(rep(2,px), rep(0,p-px))
intercept <- 1
P <- exp(intercept + X %*% beta)/(1+exp(intercept + X %*% beta))
y <- rbinom(n, size=1, prob=P)
dta <- list(y = y, X = X)
rm(y, X)

## Original function to interface
rlassologit(y ~ X, dta)

## The interface puts data as first parameter
ntbt_rlassologit(dta, y ~ X)

## so it can be used easily in a pipeline.
dta %>%
  ntbt_rlassologit(y ~ X)



## ntbt_tsls: Two-Stage Least Squares Estimation (TSLS)
## No example provided

## End(Not run)

intubate documentation built on May 2, 2019, 2:46 p.m.