Merge pull request #174 from StochasticTree/kernel-hotfix

Handle BCF models with propensity covariates in the kernel prediction functions

Author: andrewherren

R/forest.R

@@ -785,7 +785,7 @@ Forest <- R6::R6Class(
         #' Return constant leaf status of trees in a `Forest` object
         #' @return `TRUE` if leaves are constant, `FALSE` otherwise
         is_constant_leaf = function() {
-            return(is_leaf_constant_forest_container_cpp(self$forest_ptr))
+            return(is_leaf_constant_active_forest_cpp(self$forest_ptr))
         }, 
 
         #' @description

R/kernel.R

@@ -32,6 +32,7 @@
 #'
 #'   - `NULL`: It is not necessary to disambiguate when this function is called directly on a `ForestSamples` object. This is the default value of this
 #' 
+#' @param propensity (Optional) Propensities used for prediction (BCF-only).
 #' @param forest_inds (Optional) Indices of the forest sample(s) for which to compute leaf indices. If not provided, 
 #' this function will return leaf indices for every sample of a forest. 
 #' This function uses 0-indexing, so the first forest sample corresponds to `forest_num = 0`, and so on.
@@ -46,7 +47,7 @@
 #' computeForestLeafIndices(bart_model, X, "mean")
 #' computeForestLeafIndices(bart_model, X, "mean", 0)
 #' computeForestLeafIndices(bart_model, X, "mean", c(1,3,9))
-computeForestLeafIndices <- function(model_object, covariates, forest_type=NULL, forest_inds=NULL) {
+computeForestLeafIndices <- function(model_object, covariates, forest_type=NULL, propensity=NULL, forest_inds=NULL) {
     # Extract relevant forest container
     stopifnot(any(c(inherits(model_object, "bartmodel"), inherits(model_object, "bcfmodel"), inherits(model_object, "ForestSamples"))))
     model_type <- ifelse(inherits(model_object, "bartmodel"), "bart", ifelse(inherits(model_object, "bcfmodel"), "bcf", "forest_samples"))
@@ -93,6 +94,21 @@ computeForestLeafIndices <- function(model_object, covariates, forest_type=NULL,
         covariates_processed <- covariates
     }
 
+    # Handle BCF propensity covariate
+    if (model_type == "bcf") {
+        # Add propensities to covariate set if necessary
+        if (model_object$model_params$propensity_covariate != "none") {
+            if (is.null(propensity)) {
+                if (!model_object$model_params$internal_propensity_model) {
+                    stop("propensity must be provided for this model")
+                }
+                # Compute propensity score using the internal bart model
+                propensity <- rowMeans(predict(model_object$bart_propensity_model, covariates)$y_hat)
+            }
+            covariates_processed <- cbind(covariates_processed, propensity)
+        }
+    }
+    
     # Preprocess forest indices
     num_forests <- forest_container$num_samples()
     if (is.null(forest_inds)) {

man/computeForestLeafIndices.Rd

@@ -9,7 +9,7 @@
 from .forest import ForestContainer
 
 
-def compute_forest_leaf_indices(model_object: Union[BARTModel, BCFModel, ForestContainer], covariates: Union[np.array, pd.DataFrame], forest_type: str = None, forest_inds: Union[int, np.ndarray] = None):
+def compute_forest_leaf_indices(model_object: Union[BARTModel, BCFModel, ForestContainer], covariates: Union[np.array, pd.DataFrame], forest_type: str = None, propensity: np.array = None, forest_inds: Union[int, np.ndarray] = None):
     """
     Compute and return a vector representation of a forest's leaf predictions for every observation in a dataset.
 
@@ -37,6 +37,8 @@ def compute_forest_leaf_indices(model_object: Union[BARTModel, BCFModel, ForestC
             * **ForestContainer**
                 * `NULL`: It is not necessary to disambiguate when this function is called directly on a `ForestSamples` object. This is the default value of this
     
+    propensity : `np.array`, optional
+        Optional test set propensities. Must be provided if propensities were provided when the model was sampled.
     forest_inds : int or np.ndarray
         Indices of the forest sample(s) for which to compute leaf indices. If not provided, this function will return leaf indices for every sample of a forest. 
         This function uses 0-indexing, so the first forest sample corresponds to `forest_num = 0`, and so on.
@@ -88,6 +90,19 @@ def compute_forest_leaf_indices(model_object: Union[BARTModel, BCFModel, ForestC
     else:
         covariates_processed = covariates
     covariates_processed = np.asfortranarray(covariates_processed)
+
+    # Handle BCF propensity covariate
+    if model_type == "bcf":
+        if model_object.propensity_covariate != "none":
+            if propensity is None:
+                if not model_object.internal_propensity_model:
+                    raise ValueError(
+                        "Propensity scores not provided, but no propensity model was trained during sampling"
+                    )
+                propensity = np.mean(
+                    model_object.bart_propensity_model.predict(covariates), axis=1, keepdims=True
+                )
+            covariates_processed = np.c_[covariates_processed, propensity]
 
     # Preprocess forest indices
     num_forests = forest_container.num_samples()