HK2_decomposition: Implementation of Heiler & Knaus (2025) decomposition.
In MCKnaus/causalDML: Causal Double Machine Learning

HK2_decomposition

R Documentation

Implementation of Heiler & Knaus (2025) decomposition.

Implementation of Heiler & Knaus (2025) decomposition.

HK2_decomposition(
  Y,
  A,
  G,
  T_mat,
  e_mat,
  m_mat,
  sampling_weights = NULL,
  cl = NULL
)

`Y`	Numeric vector containing the outcome variable.
`A`	Aggregate treatment vector, e.g. binary. Program finds mapping to effective treatment automatically if possible.
`G`	Heterogeneity group vector. Provide as factor to control ordering, otherwise program orders treatments in ascending order or alphabetically.
`T_mat`	Logical matrix of effective treatment indicators (n x J). For example created by `prep_w_mat`.
`e_mat`	n x J matrix with propensity scores.
`m_mat`	n x J matrix with fitted outcome values.
`sampling_weights`	Optional vector of sampling weights.
`cl`	If not NULL, vector with cluster variables.

Returns an HK2_decomposition object:

`parameter`	14 x # of heterogeneity groups x # of treatment aggregates x 2 array storing point estimates and standard error of target and intermediate parameters. The ordering is c("CM","ACM","d0","s1","s2","s3","d1","d2","d3","Cov(etX,mut\|Xg)","d4","SRCT2","d4'","d5").
`IFs`	14 x # of heterogeneity groups x # of treatment aggregates x n x 6 array storing the influence fcts and its components corresponding to each parameter for further use.
`mapping`	Logical matrix storing the mapping of aggregate and effective treatment.
`label`	List of labels for further useage.
`cl`	Cluster variable if specified.