spglm: spglm Semiparametric generalized linear models for causal...

In Larsvanderlaan/causalGLM: Interpretable and robust causal inference for heterogeneous treatment effects using generalized linear models with targeted machine-learning.

spglm Semiparametric generalized linear models for causal inference Supports flexible semiparametric conditional average treatment effect (CATE), conditional odds ratio (OR), and conditional relative risk (RR) estimation Highly Adaptive Lasso (HAL) (see fit_hal), a flexible and adaptive spline regression estimator, is recommended for medium-small to large sample sizes.

Description

spglm Semiparametric generalized linear models for causal inference Supports flexible semiparametric conditional average treatment effect (CATE), conditional odds ratio (OR), and conditional relative risk (RR) estimation Highly Adaptive Lasso (HAL) (see fit_hal), a flexible and adaptive spline regression estimator, is recommended for medium-small to large sample sizes.

Usage

spglm(
  formula,
  data,
  W,
  A,
  Y,
  estimand = c("CATE", "OR", "RR"),
  learning_method = c("HAL", "SuperLearner", "glm", "glmnet", "gam", "mars", "ranger",
    "xgboost"),
  append_interaction_matrix = TRUE,
  cross_fit = FALSE,
  sl3_Learner_A = NULL,
  sl3_Learner_Y = NULL,
  wrap_in_Lrnr_glm_sp = TRUE,
  HAL_args_Y0W = list(smoothness_orders = 1, max_degree = 1, num_knots = c(10, 5, 1)),
  HAL_fit_control = list(parallel = F),
  sl3_Learner_var_Y = Lrnr_glmnet$new(family = "poisson"),
  delta_epsilon = 0.1,
  verbose = TRUE,
  warn = TRUE,
  ...
)

Arguments

formula	A R formula object specifying the parametric form of CATE, OR, or RR (depending on method).
data	A data.frame or matrix containing the numeric values corresponding with the nodes W, A and Y. Or a spglm fit object in which case previous ML fits are reused in computation. Note, only pass in a previous fit object for the same estimand and for subformulas. (See vignette)
W	A character vector of covariates contained in data
A	A character name for the treatment assignment variable contained in data
Y	A character name for the outcome variable contained in data (outcome can be continuous, nonnegative or binary depending on method)
estimand	Estimand/parameter to estimate. Choices are: 'CATE': Estimate conditional average treatment effect with Param_spCATE assuming it satisfies parametric model formula. 'OR': Estimate conditional odds ratio with Param_spOR assuming it satisfies parametric model formula. 'RR': Estimate conditional relative risk with Param_spRR assuming it satisfies parametric model formula.
learning_method	Machine-learning method to use. This is overrided if argument sl3_Learner is provided. Options are: "SuperLearner: A stacked ensemble of all of the below that utilizes cross-validation to adaptivelly choose the best learner. "HAL": Adaptive robust automatic machine-learning using the Highly Adaptive Lasso hal9001. See argumentsHAL_args_Y0W. "glm": Fit nuisances with parametric model. "glmnet": Learn using lasso with glmnet. "gam": Learn using generalized additive models with mgcv. "mars": Multivariate adaptive regression splines with earth. "ranger": Robust random-forests with the package Ranger "xgboost": Learn using a default cross-validation tuned xgboost library with max_depths 3 to 7. Note speed can vary significantly depending on learner choice!
append_interaction_matrix	Default: TRUE. This argument is passed to Lrnr_glm_semiparametric. This is a boolean for whether to estimate the conditional mean/regression of Y by combining observations with A=0,A=1 ('TRUE'), or to first E[Y|A=0,W] nonparametrically with sl3_Learner_Y or learning_method and then learning the parametric component with offsetted parametric regression ('FALSE'). If 'TRUE' the design matrix passed to the regression algorithm/learner for 'Y' is 'cbind(W,A*V)' where 'V = model.matrix(formula, as.data.frame(W))' is the design matrix specified by the argument formula. Therefore, it may not be necessary to use learners that model (treatment) interactions when this argument is TRUE. The resulting estimators are projected onto the semiparametric model, ensuring compatibility with the statistical model assumptions. In high dimensions, pool_A_when_training = FALSE may be preferred to prevent dilution of the treatment interactions in the fitting.
cross_fit	Whether to cross-fit the initial estimator. This is always set to FALSE if argument sl3_Learner is provided. learning_method = 'SuperLearner' is always cross-fitted (default). learning_method = 'xgboost' and 'ranger' are always cross-fitted regardless of the value of cross_fit All other learning_methods are only cross-fitted if 'cross_fit=TRUE'. Note, it is not necessary to cross-fit glm, glmnet, gam or mars as long as the dimension of W is not too high. In smaller samples and lower dimensions, it may fact hurt to cross-fit.
sl3_Learner_A	A sl3 Learner object to use to estimate nuisance function P(A=1|W) with machine-learning. Note, cross_fit is automatically set to FALSE if this argument is provided. If you wish to cross-fit the learner sl3_Learner then do: sl3_Learner <- Lrnr_cv$new(sl3_Learner). Cross-fitting is recommended for all tree-based algorithms like random-forests and gradient-boosting.
sl3_Learner_Y	A sl3 Learner object to use to estimate nuisance functions [Y|A=1,W] and E[Y|A=0,W] (depending on method) with machine-learning. Note, cross_fit is automatically set to FALSE if this argument is provided. Keep in mind the value of the argument pool_A_when_training. If FALSE then E[Y|A=0,W] is estimated by itself. Therefore, it may not be needed to add interactions, since treatment interactions are automatic by stratification. If TRUE, the design matrix passed to the pooled learner contains A*V where V is the design matrix obtained from formula. For some learners, it may also be unnecessary to include interactions in this case. #' If you wish to cross-fit the learner sl3_Learner then do: sl3_Learner <- Lrnr_cv$new(sl3_Learner). Cross-fitting is recommended for all tree-based algorithms like random-forests and gradient-boosting.
wrap_in_Lrnr_glm_sp	Mostly for internal use (should be TRUE usually). Whether sl3_Learner_Y should be wrapped in a Lrnr_glm_semiparametric object.
HAL_args_Y0W	A list of parameters for the semiparametric Highly Adaptive Lasso estimator for E[Y|A=0,W]. Possible parameters are: 1. 'smoothness_orders': Smoothness order for HAL estimator of E[Y|A=0,W] (see fit_hal) smoothness_order_Y0W = 1 is piece-wise linear. smoothness_order_Y0W = 0 is piece-wise constant. 2. 'max_degree': Max interaction degree for HAL estimator of E[Y|A=0,W] (see fit_hal) 3. 'num_knots': A vector of the number of knots by interaction degree for HAL estimator of E[Y|A=0,W] (see fit_hal). Used to generate spline basis functions.
HAL_fit_control	See the argument 'fit_control' of (see fit_hal).
sl3_Learner_var_Y	A sl3-Learner for the conditional variance of 'Y'. Only used if 'estimand = "CATE"' and by default is estimated using Poisson-link LASSO regression with 'Lrnr_glmnet'. If conditional variance is constant, set 'sl3_Learner_var_Y = Lrnr_mean$new()'.
delta_epsilon	Step size of iterative targeted maximum likelihood estimator. 'delta_epsilon = 1 ' leads to large step sizes and fast convergence. 'delta_epsilon = 0.005' leads to slower convergence but possibly better performance. Useful to set to a large value in high dimensions.
...	Not used

Larsvanderlaan/causalGLM documentation built on April 14, 2022, 12:51 a.m.