CATEs

Coverage

The simulations are based on the the make_heterogeneous_data-DGP with \(2000\) observations. The groups are defined based on the first covariate, analogously to the CATE IRM Example, but rely on LightGBM to estimate nuisance elements (due to time constraints).

The non-uniform results (coverage, ci length and bias) refer to averaged values over all groups (point-wise confidende intervals).

DoubleML Version                            0.11.5.dev55
Script                         IRMCATECoverageSimulation
Date                                    2026-08-10 11:53
Total Runtime (minutes)                          80.2954
Python Version                                    3.12.3
Config File              scripts/irm/irm_cate_config.yml
ⓘLearner g Learner m Bias CI Length Coverage Uniform CI Length Uniform Coverage
LGBM Regr. LGBM Clas. 0.240 1.244 0.965 1.755 0.971
LGBM Regr. Logistic 0.111 0.547 0.946 0.773 0.941
Linear LGBM Clas. 0.250 1.238 0.954 1.750 0.955
Linear Logistic 0.114 0.567 0.949 0.803 0.943
Linear RF Clas. 0.120 0.605 0.953 0.856 0.941
RF Regr. Logistic 0.110 0.547 0.948 0.773 0.928
RF Regr. RF Clas. 0.117 0.589 0.952 0.830 0.931
Coverage for 95%-Confidence Interval over 1000 Repetitions
ⓘLearner g Learner m Bias CI Length Coverage Uniform CI Length Uniform Coverage
LGBM Regr. LGBM Clas. 0.240 1.044 0.924 1.756 0.970
LGBM Regr. Logistic 0.111 0.459 0.898 0.772 0.941
Linear LGBM Clas. 0.250 1.039 0.905 1.750 0.953
Linear Logistic 0.114 0.476 0.897 0.804 0.937
Linear RF Clas. 0.120 0.508 0.904 0.857 0.939
RF Regr. Logistic 0.110 0.459 0.899 0.773 0.931
RF Regr. RF Clas. 0.117 0.494 0.903 0.830 0.930
Coverage for 90%-Confidence Interval over 1000 Repetitions