Share your thoughts, 1 month free Claude Pro on usSee more

Causal effect estimation

Benchmarks

Dataset Name	SOTA Method	Metric
Flickr (in-sample)	C-VAE	Epsilon ATE0.6	21	2mo ago
BlogCatalog simulated (In-sample)	C-HDNet	ATE Error (Epsilon)0.6	21	2mo ago
Flickr 1.0 (out-of-sample)	C-HDNet	Epsilon ATE0.5	21	2mo ago
BlogCatalog (out-of-sample)	C-HDNet	ϵATE0.3	21	2mo ago
Synthetic Binary Treatment, d=200	DDML-MLP	MAE0.1942	15	8d ago
Synthetic Binary Treatment, d=100		MAE0.1432	15	8d ago
Synthetic Binary Treatment, d=50	DDML-RF	MAE0.052	15	8d ago
Synthetic Binary Treatment, d=20	DDML-MLP	MAE0.0488	15	8d ago
BlogCatalog (BC) Cosine kernel (out-of-sample)	Proposed_est	ADE0.0004	15	22d ago
6-node graph n=5000 samples, averaged over 100 trials	IIC	Absolute Bias0	12	1mo ago
180 synthetic causal datasets (test)		\|ATE\| Error0.406	11	3mo ago
IHDP U -> X (test)		Mean Absolute ATE Prop Error53.4	11	3mo ago
BlogCatalog (BC) RBF kernel (out-of-sample)	GDML	ADE0	10	22d ago
ACIC out-of-sample 2018		Top-1 % PEHE Error0	10	26d ago
ACIC in-sample 2018	Repflow	TOP1 % (PEHE sqrt)37.5	10	26d ago
IHDP (sub-sampled)	GP-CATE	Coverage96	9	6d ago
IHDP 40 confounded datasets with no true instrument candidate	ZNet	Mean Abs ATE Prop Error0.566	9	3mo ago
IHDP 1000 realizations from NPCI package (heldout)	baseline + t-reg	MAE0.2	9	3mo ago
IHDP 1000 realizations from NPCI package (train val)	Dragonnet	MAE0.14	9	3mo ago
CausalSim-Diabetes	SVAR-FM	Sign Correct100	8	21d ago
Synthetic Continuous 200	DDML-RF	MAE0.2455	7	8d ago
Synthetic Continuous 100	DDML-MLP	Mean Absolute Error0.2226	7	8d ago
Synthetic Continuous 50	DDML-RF	MAE0.0706	7	8d ago
Synthetic Continuous 20	DDML-MLP	MAE0.082	7	8d ago
Pubmed	Ma & Tresp	ADE0.02	7	3mo ago

Showing 25 of 62 rows