Datasets:

Bauxitiego
/

uplift-modeling-synthetic-benchmark

x0 float64	x1 float64	x2 float64	x3 float64	x4 float64	x5 float64	treatment int64	outcome float64	true_cate float64
0.345584	0.821618	0.330437	-1.303157	0.905356	0.446375	1	1.92436	1.222931
-0.536953	0.581118	0.364572	0.294132	0.028422	0.546713	0	-0.258311	0.007274
-0.736454	-0.16291	-0.482119	0.598846	0.039722	-0.292457	1	-0.589083	-0.351047
-0.781908	-0.257192	0.008142	-0.275603	1.294064	1.006724	1	-1.032591	-0.402345
-2.711162	-1.889013	-0.174772	-0.42219	0.213643	0.217322	1	0.027582	-1.066117
2.117839	-1.112021	-0.377605	2.042772	0.646703	0.663063	1	2.478413	1.162919
-0.514006	-1.648075	0.167465	0.109014	-1.227352	-0.683227	0	1.044255	-0.641939
-0.072044	-0.944752	-0.09827	0.095483	0.035586	-0.506292	1	1.262651	0.108928
0.593748	0.891167	0.320848	-0.81823	0.731652	-0.50144	0	0.71158	1.479051
0.879161	-1.071787	0.914467	-0.020063	-1.248749	-0.313899	1	3.318461	1.044934
0.054102	0.272791	-0.982188	-1.107373	0.199585	-0.46675	1	1.795186	0.662813
0.235506	0.75952	-1.648787	0.254388	1.224647	-0.297527	1	2.156859	1.067118
-0.810815	0.752244	0.253447	0.895883	-0.345216	-1.481818	0	-0.187931	-0.112863
-0.110011	-0.445828	0.775324	0.193633	-1.630849	-1.195163	0	0.040679	0.202717
0.883789	0.679765	-0.640243	-0.001049	0.445574	0.468404	1	2.893403	1.57212
0.876242	0.256486	-0.094828	-0.258848	1.055743	-2.250854	1	2.412098	1.442321
-0.138655	0.033	-1.425349	0.332814	-0.651281	0.862445	0	0.511876	0.304865
-0.125592	0.669153	1.218844	0.38293	-0.875721	-1.514319	1	0.882551	0.514555
1.753384	-0.111292	-0.688565	0.144257	-0.191411	0.852142	1	2.070505	1.456277
0.033928	0.01375	-0.71458	0.469568	-1.033867	0.665889	0	0.282629	0.554973
1.523938	-1.524686	-2.466229	0.616879	2.547898	-1.000925	0	1.972752	1.022127
-1.250696	0.588969	-0.840722	-0.506025	-0.348117	0.532002	0	-1.148006	-0.277448
-0.405302	0.277883	-0.176533	-0.844671	-0.319826	-0.9504	1	-0.226064	0.04068
0.006515	-1.123866	-1.092894	1.456962	-0.053184	-0.053902	0	1.550312	0.172612
0.511536	-0.420857	-0.228535	0.425149	0.282416	-1.159297	1	1.273181	1.019102
0.833343	-0.590435	-1.056079	-0.900475	-0.390545	1.6273	1	1.721836	1.171157
-1.175536	0.160076	-2.137824	-0.001567	0.899566	-0.236663	0	-0.999823	-0.394848
-0.629355	0.231511	0.700152	0.663658	1.972474	0.209167	0	-1.24038	-0.167616
-0.59241	-0.125979	-0.072499	0.108737	-0.030028	0.173966	0	-0.975162	-0.248503
-1.67085	0.829629	-0.574739	-1.173159	0.637751	1.317326	1	-1.417866	-0.237891
0.493028	0.161159	-0.93222	2.871567	0.880259	-1.139295	1	1.496062	1.177216
-0.779638	0.086979	-1.554731	0.16863	-0.459072	1.226271	0	-1.363996	-0.298004
0.962155	-2.711285	0.041703	-1.617467	1.109638	0.168106	1	5.758335	0.580961
0.548405	-1.065125	1.82843	2.020073	-1.064771	0.372815	0	1.549594	0.85695
-0.673302	-0.02357	-1.265637	1.867146	-0.96918	-0.296084	0	-0.055725	-0.272814
0.501483	-0.647561	-0.239312	-0.56364	-0.133461	-1.170543	0	0.647608	0.942206
-0.437988	-0.206893	-0.333726	0.05669	-0.293102	0.753211	0	-0.445537	-0.138419
-0.323196	-0.13665	-0.664781	-0.526515	-1.264493	0.518785	0	0.549908	0.00893
-1.142518	-0.745856	0.359245	0.402573	-0.400115	-2.019266	1	-1.20679	-0.660866
0.420513	0.259563	-1.412381	0.770322	-0.7011	-1.126188	0	0.771525	1.136451
0.095731	-0.17847	0.202624	-1.605748	1.81223	-0.602659	0	-0.05956	0.589076
-1.539659	0.618842	-0.354804	0.324858	-0.339608	-0.05974	1	-0.518787	-0.294814
0.245773	-0.746653	0.67874	-0.4699	-0.869687	0.077032	0	1.055894	0.628823
0.445041	-0.229079	-0.86252	0.619786	-1.760329	-1.030864	1	2.55094	1.01465
0.039523	-1.361059	0.027994	-0.054863	0.89874	-0.91479	0	1.384129	0.150897
-0.625907	0.333182	-2.457564	3.100042	-0.698651	-0.729835	1	-1.038814	-0.134743
0.861128	-0.039832	-1.779429	0.626927	0.855378	-0.449946	0	0.818328	1.347619
-0.2816	0.485985	-0.90878	0.438389	0.199299	-0.674933	0	0.036295	0.246845
-1.392102	-0.225606	-0.875422	1.00141	0.144085	0.782085	1	-1.36109	-0.537436
0.134622	0.262901	-0.782999	0.668047	1.784698	-0.309688	0	0.509571	0.778103
-0.592775	-0.157837	-0.48128	-0.701479	0.138194	-0.290918	1	-1.074371	-0.25833
1.438874	0.000202	0.323912	0.952022	-0.300756	1.436737	1	1.781829	1.473722
-0.632694	-0.808328	-0.366267	-0.114717	-1.401318	-0.035095	0	-0.010904	-0.481847
-1.667487	1.392138	-0.080997	-0.641958	-0.908338	-0.38443	1	0.290003	-0.069005
-0.223079	-1.044511	-0.919795	-0.187175	-0.520784	0.939213	0	0.492085	-0.136018
1.13787	0.016021	0.4736	-1.335188	0.637418	-0.030583	0	0.490601	1.441061
0.484668	1.600356	-2.280858	0.260948	-1.099119	0.592197	1	3.238579	1.601335
-1.313313	-0.4954	0.202734	0.613509	0.074794	-0.792831	0	-0.373943	-0.610463
-0.553532	0.884503	-0.0055	-1.683814	0.843663	0.416248	1	-0.039457	0.084715
0.873416	-0.336627	0.828157	-1.061065	0.57	-0.49038	0	1.739492	1.263318
0.674352	1.005644	-0.73599	-0.051229	0.038955	1.189665	1	2.570439	1.568086
0.710558	-1.219278	0.457608	0.745089	2.123806	-1.679149	1	2.223632	0.922104
-0.536352	1.333371	-1.35507	-1.199461	0.517082	1.018409	0	1.070725	0.233451
-0.66868	0.540127	0.116955	1.518749	-0.001518	0.990247	0	-1.22569	-0.100822
-0.903118	-0.184879	-0.096704	1.139108	0.579613	-0.751753	1	-1.682307	-0.430616
0.681968	0.770631	-0.111646	-0.257662	-0.193803	-1.694992	1	2.625062	1.502255
0.188743	0.234578	-0.865529	0.742446	-1.372967	-0.551813	1	0.950935	0.846159
-0.475219	1.98751	-1.599184	0.56263	0.942123	0.377726	0	2.07949	0.483805
1.181808	-1.002738	-2.280259	0.766818	-1.195828	-0.3264	1	2.181731	1.143085
-1.178846	1.052342	0.86515	-0.720064	0.903492	0.121277	0	-0.951864	-0.127717
-0.139389	0.057417	-0.202749	0.615005	0.310348	-0.349467	0	0.043975	0.311136
1.006256	-0.611476	0.286934	0.422155	1.478807	-0.505441	0	0.538899	1.223387
1.743082	0.176165	-0.193977	-0.672473	0.586293	0.048935	1	1.511479	1.542191
-1.103609	-1.127887	-0.555519	-0.676006	1.090302	1.375538	1	-0.146328	-0.767963
0.86856	0.357869	-0.428238	0.054483	0.884707	2.130972	1	1.706126	1.469813
0.912899	-0.280412	0.038025	-0.482545	-0.782399	-0.184242	0	0.826421	1.29442
0.195615	1.835902	0.052032	1.357358	1.767489	0.078808	1	2.982785	1.336052
1.607799	0.714935	-0.417535	0.26518	0.022878	-0.232339	1	2.90539	1.69853
-0.198781	0.144209	0.434181	-0.884882	-0.009388	-1.522355	0	0.177205	0.253624
0.260795	0.623599	0.163955	0.285853	0.589943	-0.665101	0	-0.236246	1.059468
-0.239049	0.510922	1.001886	0.394921	2.552859	-0.091361	0	-0.948598	0.30932
0.99966	1.266728	-0.13377	-0.820513	-1.182603	0.162277	0	1.367029	1.785074
1.110292	0.271668	0.172856	-0.379214	0.56103	-2.135831	1	3.259474	1.512446
0.232373	0.028126	-1.37034	2.175598	-1.387413	-1.07752	0	0.341462	0.843536
-1.200863	1.110368	-0.888085	0.668656	0.58751	0.25967	1	0.96389	-0.11383
-1.307579	-0.612106	1.673115	-1.290754	-0.831655	-0.162247	0	-1.627669	-0.644826
0.80899	0.251639	0.742044	-1.067294	0.944777	0.569029	1	2.066535	1.413214
-1.594565	1.539916	2.292398	-0.768577	0.055653	1.397423	0	0.381868	-0.021434
-1.48124	-1.98986	-1.296944	-0.567067	-0.578442	0.607347	0	1.742268	-1.073727
0.268536	-1.246077	0.56738	1.874081	1.197054	0.991107	1	2.013842	0.508523
0.020178	0.983631	-0.966133	0.751015	-0.087248	1.130906	0	1.101481	0.825347
0.466011	-1.08988	0.127993	1.218221	-1.129234	-0.55914	1	1.528251	0.776779
-0.768673	-1.495646	0.961268	1.310997	0.800043	0.24257	0	-0.064158	-0.767442
-0.052031	0.23938	-0.731556	0.893846	1.045492	0.932544	1	0.300123	0.493925
-0.530948	0.078404	-0.165781	1.788575	0.179696	-1.882071	0	-0.077626	-0.138511
0.398506	1.882266	0.685176	0.87948	0.036061	-1.971389	0	2.033871	1.600133
-1.810258	-1.245948	-0.126832	0.309998	0.689414	-0.340458	1	-0.112122	-0.865063
0.956929	-0.279749	-0.70449	0.851642	-0.914445	-2.729085	0	1.194314	1.308842
-1.059957	0.09416	-3.080178	-0.357138	-0.33209	-1.426456	1	-1.415445	-0.391892
-1.482503	-0.462462	-0.5471	1.259919	0.374863	-1.580855	0	-0.768305	-0.615594

End of preview. Expand in Data Studio

Synthetic uplift benchmark with known ground-truth treatment effect

100,000 training rows, 20,000 validation rows, generated for uplift-modeling's Gate 0: checking that meta-learners (S/T/X-learner) actually recover a real treatment effect before trusting them on data where no individual ground truth is ever available - which is true of essentially all real causal-inference data, by the fundamental problem of causal inference (nobody observes both potential outcomes for the same unit).

This dataset exists specifically because that real ground truth is missing everywhere else. true_cate is the exact, closed-form conditional average treatment effect used to generate each row - not an estimate, the actual number. Useful for validating any uplift-modeling implementation the same way Gate 0 validated the ones in that repo, not just this project's.

Columns

column	meaning
`x0`-`x5`	six covariates, standard normal. Only `x0` and `x1` drive the treatment effect; `x2`-`x5` are pure noise a learner has to learn to ignore.
`treatment`	0 or 1, assigned by a fair coin independent of the covariates - literally randomized.
`outcome`	observed outcome: baseline (nonlinear in x0, x1) plus `treatment * true_cate` plus noise.
`true_cate`	the exact treatment effect for this row, a smooth sigmoid function of x0 plus a linear term in x1. Never available on real data - this is the entire point of the dataset.

Why the effect is smooth, not a step function

A discontinuous treatment effect (e.g. positive above a threshold, negative below) is trivially recoverable by a tree-based T-learner and wouldn't test anything real. The sigmoid keeps the effect boundary soft, so recovering it actually requires the learner to model a continuous function of the covariates, not just find a split point.

Generation code

src/synthetic.py in the parent repo, generate(n, seed). Train uses seed 1, validation uses seed 2 - different seeds, not a held-out slice of the same draw, so there's no leakage in the usual sense (both are i.i.d. samples from the same generating process) but the split is real and reproducible.

License

Apache 2.0. Fully synthetic - no real user, ad, or personal data of any kind.

Downloads last month: 43