SSM with Missingness at Random

ATE Coverage

The simulations are based on the the make_ssm_data-DGP with \(500\) observations. The simulation considers data under missingness at random.

DoubleML Version                           0.9.0
Script                   ssm_mar_ate_coverage.py
Date                         2024-09-09 09:44:28
Total Runtime (seconds)              8451.345644
Python Version                            3.12.5
Table 1: Coverage for 95.0%-Confidence Interval over 1000 Repetitions
Learner g Learner m Learner pi Bias CI Length Coverage
LGBM LGBM LGBM 1.525 7.024 0.981
LGBM LGBM Logistic 0.615 3.076 0.973
LGBM Logistic LGBM 0.654 3.069 0.985
LGBM Logistic Logistic 0.127 0.643 0.958
Lasso LGBM LGBM 1.270 5.995 0.981
Lasso LGBM Logistic 0.622 2.790 0.955
Lasso Logistic LGBM 0.613 2.740 0.970
Lasso Logistic Logistic 0.123 0.610 0.961
Table 2: Coverage for 90.0%-Confidence Interval over 1000 Repetitions
Learner g Learner m Learner pi Bias CI Length Coverage
LGBM LGBM LGBM 1.525 5.895 0.934
LGBM LGBM Logistic 0.615 2.582 0.927
LGBM Logistic LGBM 0.654 2.576 0.945
LGBM Logistic Logistic 0.127 0.540 0.914
Lasso LGBM LGBM 1.270 5.031 0.939
Lasso LGBM Logistic 0.622 2.341 0.887
Lasso Logistic LGBM 0.613 2.300 0.919
Lasso Logistic Logistic 0.123 0.512 0.897