Merge pull request #99 from StochasticTree/kernel_updates

Simplifying and streamlining the forest kernel interface

Author: andrewherren

DESCRIPTION

@@ -17,17 +17,17 @@ RoxygenNote: 7.3.1
 LinkingTo: 
     cpp11
 Suggests: 
+    doParallel, 
+    foreach, 
+    ggplot2,
     knitr,
-    rmarkdown,
+    latex2exp,
     Matrix,
-    tgp,
     MASS,
     mvtnorm,
-    ggplot2,
-    latex2exp,
+    rmarkdown,
     testthat (>= 3.0.0), 
-    foreach, 
-    doParallel
+    tgp
 VignetteBuilder: knitr
 SystemRequirements: C++17
 Imports: 

NAMESPACE

@@ -7,8 +7,8 @@ S3method(predict,bcf)
 export(bart)
 export(bcf)
 export(calibrate_inverse_gamma_error_variance)
-export(computeForestKernels)
 export(computeForestLeafIndices)
+export(computeMaxLeafIndex)
 export(convertBARTModelToJson)
 export(convertBCFModelToJson)
 export(createBARTModelFromCombinedJson)
@@ -26,7 +26,6 @@ export(createForestContainer)
 export(createForestCovariates)
 export(createForestCovariatesFromMetadata)
 export(createForestDataset)
-export(createForestKernel)
 export(createForestModel)
 export(createOutcome)
 export(createRNG)

R/cpp11.R

@@ -340,32 +340,12 @@ predict_forest_raw_single_forest_cpp <- function(forest_samples, dataset, forest
   .Call(`_stochtree_predict_forest_raw_single_forest_cpp`, forest_samples, dataset, forest_num)
 }
 
-forest_kernel_cpp <- function() {
-  .Call(`_stochtree_forest_kernel_cpp`)
+forest_container_get_max_leaf_index_cpp <- function(forest_container, forest_num) {
+  .Call(`_stochtree_forest_container_get_max_leaf_index_cpp`, forest_container, forest_num)
 }
 
-forest_kernel_compute_leaf_indices_train_cpp <- function(forest_kernel, covariates_train, forest_container, forest_num) {
-  invisible(.Call(`_stochtree_forest_kernel_compute_leaf_indices_train_cpp`, forest_kernel, covariates_train, forest_container, forest_num))
-}
-
-forest_kernel_compute_leaf_indices_train_test_cpp <- function(forest_kernel, covariates_train, covariates_test, forest_container, forest_num) {
-  invisible(.Call(`_stochtree_forest_kernel_compute_leaf_indices_train_test_cpp`, forest_kernel, covariates_train, covariates_test, forest_container, forest_num))
-}
-
-forest_kernel_get_train_leaf_indices_cpp <- function(forest_kernel) {
-  .Call(`_stochtree_forest_kernel_get_train_leaf_indices_cpp`, forest_kernel)
-}
-
-forest_kernel_get_test_leaf_indices_cpp <- function(forest_kernel) {
-  .Call(`_stochtree_forest_kernel_get_test_leaf_indices_cpp`, forest_kernel)
-}
-
-forest_kernel_compute_kernel_train_cpp <- function(forest_kernel, covariates_train, forest_container, forest_num) {
-  .Call(`_stochtree_forest_kernel_compute_kernel_train_cpp`, forest_kernel, covariates_train, forest_container, forest_num)
-}
-
-forest_kernel_compute_kernel_train_test_cpp <- function(forest_kernel, covariates_train, covariates_test, forest_container, forest_num) {
-  .Call(`_stochtree_forest_kernel_compute_kernel_train_test_cpp`, forest_kernel, covariates_train, covariates_test, forest_container, forest_num)
+compute_leaf_indices_cpp <- function(forest_container, covariates, forest_nums) {
+  .Call(`_stochtree_compute_leaf_indices_cpp`, forest_container, covariates, forest_nums)
 }
 
 sample_gfr_one_iteration_cpp <- function(data, residual, forest_samples, tracker, split_prior, rng, feature_types, cutpoint_grid_size, leaf_model_scale_input, variable_weights, a_forest, b_forest, global_variance, leaf_model_int, pre_initialized) {

R/kernel.R

@@ -81,13 +81,13 @@ reference:
   - createForestModel
   - ForestSamples
   - createForestContainer
-  - ForestKernel
-  - createForestKernel
   - CppRNG
   - createRNG
   - calibrate_inverse_gamma_error_variance
   - preprocessBartParams
   - preprocessBcfParams
+  - computeForestLeafIndices
+  - computeMaxLeafIndex
 
 - subtitle: Random Effects
   desc: >
@@ -105,8 +105,6 @@ reference:
   - sample_sigma2_one_iteration
   - sample_tau_one_iteration
   - sample_tau_one_iteration
-  - computeForestKernels
-  - computeForestLeafIndices
 
 - title: Package info
   desc: >

_pkgdown.yml

@@ -36,6 +36,9 @@ class ForestContainer {
   void PredictInPlace(ForestDataset& dataset, std::vector<double>& output);
   void PredictRawInPlace(ForestDataset& dataset, std::vector<double>& output);
   void PredictRawInPlace(ForestDataset& dataset, int forest_num, std::vector<double>& output);
+  void PredictLeafIndicesInplace(Eigen::Map<Eigen::Matrix<double, Eigen::Dynamic, Eigen::Dynamic, Eigen::ColMajor>>& covariates, 
+                                 Eigen::Map<Eigen::Matrix<int, Eigen::Dynamic, Eigen::Dynamic, Eigen::ColMajor>>& output, 
+                                 std::vector<int>& forest_indices, int num_trees, data_size_t n);
 
   inline TreeEnsemble* GetEnsemble(int i) {return forests_[i].get();}
   inline int32_t NumSamples() {return num_samples_;}

include/stochtree/container.h

@@ -241,6 +241,20 @@ class TreeEnsemble {
     }
   }
 
+  /*!
+   * \brief Obtain a 0-based "maximum" leaf index for an ensemble, which is equivalent to the sum of the 
+   *        number of leaves in each tree. This is used in conjunction with `PredictLeafIndicesInplace`,
+   *        which returns an observation-specific leaf index for every observation-tree pair.
+   */
+  int GetMaxLeafIndex() {
+    int max_leaf = 0;
+    for (int j = 0; j < num_trees_; j++) {
+      auto &tree = *trees_[j];
+      max_leaf += tree.NumLeaves();
+    }
+    return max_leaf;
+  }
+
   /*!
    * \brief Obtain a 0-based leaf index for every tree in an ensemble and for each 
    *        observation in a ForestDataset. Internally, trees are stored as essentially 
@@ -274,7 +288,7 @@ class TreeEnsemble {
    *
    *        Note: this assumes the creation of a vector of column indices of size 
    *        `dataset.NumObservations()` x `ensemble.NumTrees()`
-   * \param ForestDataset Dataset with which to predict leaf indices from the tree
+   * \param covariates Matrix of covariates
    * \param output Vector of length num_trees*n which stores the leaf node prediction
    * \param num_trees Number of trees in an ensemble
    * \param n Size of dataset
@@ -292,6 +306,39 @@ class TreeEnsemble {
     }
   }
 
+  /*!
+   * \brief Obtain a 0-based leaf index for every tree in an ensemble and for each 
+   *        observation in a ForestDataset. Internally, trees are stored as essentially 
+   *        vectors of node information, and the leaves_ vector gives us node IDs for every 
+   *        leaf in the tree. Here, we would like to know, for every observation in a dataset, 
+   *        which leaf number it is mapped to. Since the leaf numbers themselves 
+   *        do not carry any information, we renumber them from 0 to `leaves_.size()-1`. 
+   *        We compute this at the tree-level and coordinate this computation at the 
+   *        ensemble level.
+   *
+   *        Note: this assumes the creation of a matrix of column indices with `num_trees*n` rows
+   *        and as many columns as forests that were requested from R / Python
+   * \param covariates Matrix of covariates
+   * \param output Matrix with num_trees*n rows and as many columns as forests that were requested from R / Python
+   * \param column_ind Index of column in `output` into which the result should be unpacked
+   * \param num_trees Number of trees in an ensemble
+   * \param n Size of dataset
+   */
+  void PredictLeafIndicesInplace(Eigen::Map<Eigen::Matrix<double, Eigen::Dynamic, Eigen::Dynamic, Eigen::ColMajor>>& covariates, 
+                                 Eigen::Map<Eigen::Matrix<int, Eigen::Dynamic, Eigen::Dynamic, Eigen::ColMajor>>& output, 
+                                 int column_ind, int num_trees, data_size_t n) {
+    CHECK_GE(output.size(), num_trees*n);
+    int offset = 0;
+    int max_leaf = 0;
+    for (int j = 0; j < num_trees; j++) {
+      auto &tree = *trees_[j];
+      int num_leaves = tree.NumLeaves();
+      tree.PredictLeafIndexInplace(covariates, output, column_ind, offset, max_leaf);
+      offset += n;
+      max_leaf += num_leaves;
+    }
+  }
+
   /*!
    * \brief Obtain a 0-based leaf index for every tree in an ensemble and for each 
    *        observation in a ForestDataset. Internally, trees are stored as essentially