Pith. sign in

REVIEW 4 major objections 4 minor 46 references

Deep Learning of Continuous and Structured Policies for Aggregated Heterogeneous Treatment Effects

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims a Neural Augmented Naive Bayes layer estimates conditional expected outcomes under continuous treatment, enabling direct ranking of aggregated heterogeneous treatment effects with arbitrary treatment policies.

desk verdict Novel neural stacking for multi-factor ranking, but Eq. 2 misidentifies a counterfactual expectation; the causal claims don't hold up. read the letter →

arxiv 2507.05511 v1 pith:Y3NYDRXL submitted 2025-07-07 cs.LG stat.ME

Jennifer Y. Zhang , Shuyang Du , Will Y. Zou This is my paper
classification cs.LGstat.ME
keywords HeterogeneoustreatmenteffectupliftmodelingcontinuousneuralaugmentednaiveBayesdirectrankingcost-awareoptimizationcausalinferencestructuredpolicy
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to extend heterogeneous treatment effect (HTE) estimation and uplift ranking from a single binary treatment to a structured treatment policy that can include continuous treatment intensities and discrete treatment assignments. To do this it introduces the Neural Augmented Naive Bayes Layer (NANBL), a factored neural layer that represents the conditional expected outcome under continuous treatment as a normalized weighted sum over the observed treated cohort. The authors claim this layer makes direct ranking of aggregated treatment effects tractable with an arbitrary number of treatment factors, all trainable end-to-end in a single deep model. They report improved uplift ranking metrics on Ponpare, US Census 1990, and Covertype datasets relative to several baselines. A sympathetic reader would therefore take the paper's contribution to be a flexible and scalable machinery for structured treatment policies, with the caveat that the key identification step is asserted rather than proven.

What carries the argument

The Neural Augmented Naive Bayes Layer (NANBL) is the machinery that carries the argument. It replaces the intractable partition function of a Bayesian decomposition with normalized products of neural network outputs: the prior $f(x)$ and likelihood $g(x,\rho_c)$ are forward functions of MLPs, with $g$ shaped as a sigmoid-derivative bell centered at $\hat{g}(x)$. The layer outputs normalized effectiveness probabilities $p(I_x|\rho_c,x)$ (Eq. 4), and Algorithm 1 recurses over treatment variables so that each additional factor adds another NANBL layer. This factored construction is what lets the model handle continuous intensity and discrete assignment jointly, and what makes the whole objective differentiable for end-to-end optimization.

What would settle it

Simulate a population with known potential outcomes $Y_\rho$ for a range of doses $\rho$, assign treatment with a propensity that depends on covariate $x$ (so unconfoundedness does not hold), and compare the NANBL estimate from Eq. (2) with the true $E[Y_\rho|x]$ at held-out test points. If the weighted sum over treated units diverges from the true counterfactual mean under such confounding, the ranking by uplift would also diverge from the oracle ranking; this would settle whether the identification step holds.

Watch

Extended reading notes

Core claim

The central claim is that the conditional expected outcome under a continuous treatment, $E[Y_{\rho_c}|x]$ in Eq. (1), can be estimated by the NANBL expression in Eq. (2): a sum over the treated cohort of observed outcomes weighted by posterior probabilities $p(I_x|\rho_c,x)$, where the posterior is decomposed by Bayes rule into a neural prior $f(x)$ and a bell-shaped neural likelihood $g(x,\rho_c)=\sigma(\rho_c-\hat{g}(x))(1-\sigma(\rho_c-\hat{g}(x)))$. Under the Naive Bayes assumption $I_x\perp\rho_c|x$, this yields Eq. (4), a differentiable normalized layer that can be stacked recursively for discrete treatment assignment and further factors. The paper argues that this construction gives a valid estimator of the conditional counterfactual outcome and hence of the average treatment effect $\tau^*(x|\rho_c)$, enabling direct ranking of subjects by aggregated treatment-effect functions such as the ratio of incremental value to incremental cost.

Load-bearing premise

The load-bearing premise is that the normalized weighted sum over the treated cohort in Eq. (2), with weights summing to one, equals the counterfactual conditional expectation $E[Y_{\rho_c}|x]$ for the whole population; the paper does not justify this identification.

Editorial extensions

If this is right

  • With NANBL stacked for arbitrary factors, uplift ranking can in principle accommodate any structured policy $\Pi_x=(\rho_c,t_a,\ldots)$ in one jointly optimized deep model.
  • Direct ranking of aggregated treatment effects, such as the ROI objective in Eq. (6), becomes differentiable and trainable end-to-end.
  • Propensity weighting can be folded into the ranking objective (Eq. 10), and the paper reports that it improves test-set generalization of both Direct Ranking and R-learner on Census and Covertype.
  • Barrier functions can enforce percentage or budget constraints on the fraction of subjects treated while keeping the objective smooth.
  • On the public datasets tested, SCPM reports higher AUUC, AUQC, LIFT@30, and AUCC than the compared baselines, with the caveat that the marginal gains vary by metric and dataset.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct extension the authors leave implicit is that the same layer could rank under cost constraints jointly with treatment intensity, so a deployment could optimize discount size per user rather than only ranking a fixed policy.
  • If the identification step were repaired with proper inverse-propensity weighting inside Eq. (2), the architecture would likely retain its scalability while gaining a rigorous causal justification; the current empirical gains may partly reflect propensity weighting and regularization rather than the NANBL decomposition itself.
  • The bell-shaped likelihood $g(x,\rho_c)$ assumes treatment intensity is unimodally centered on $\hat{g}(x)$; a testable extension would replace it with a mixture or non-parametric density to handle multi-modal dose distributions, and check whether ranking quality changes.
  • Because the paper's test-time ranker uses only the prior network $f(x)$, an ablation that removes the NANBL likelihood terms would isolate how much of the reported gain comes from the factored layer versus the underlying deep regressor.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper proposes a Neural Augmented Naive Bayes Layer (NANBL) and a Structured and Continuous Policy Model (SCPM) for estimating and ranking heterogeneous treatment effects when treatments are continuous or bundled into structured policies. The central identification step, Eq. (2), rewrites the conditional potential-outcome mean E[Y_{rho_c}|x] as a normalized weighted average of observed outcomes from the treated cohort, with weights given by a learned posterior. The paper then derives a recursive stacking procedure for multiple treatment factors, formulates a cost-aware ratio objective, and reports experiments on the Ponpare, US Census, and Covertype datasets.

Significance. If the central identification were valid, the NANBL/SCPM framework would be a useful extension of uplift modeling to continuous and multi-factor treatments, and the release of preprocessed Ponpare data and code would be a service to the community. However, the central estimator is not justified as a counterfactual expectation, the main objective omits the propensity weighting that the paper itself says is necessary, and the empirical support is weaker than claimed. The contribution therefore does not currently establish an advance over existing causal-effect estimators.

major comments (4)
  1. [Section 3.2, Eq. (2)] The equality E[Y_rho_c|x] = sum_{x in cohort, rho_c neq 0} p(I_x | rho_c, x) Y_rho_c is asserted without identification assumptions. The right-hand side is a normalized weighted average over subjects who actually received some nonzero dose; if the sum ranges over subjects with different observed doses, it marginalizes over the observed dose distribution and cannot represent the conditional expectation at a fixed dose rho_c unless all observed doses coincide. If instead the sum is meant to condition on the observed dose, it estimates at best E[Y | T=1, x] under ignorability, not E[Y_rho_c | x] for the full population. The paper does not invoke unconfoundedness, overlap, or inverse propensity weighting at this step, and Section 3.5's later statement that propensity is needed for non-random treatment confirms the omission.
  2. [Section 3.4, Eqs. (7)-(9)] The SCPM training objective maximizes a ratio in which the same learned probabilities p_i(I_x | Pi_x, theta, x) multiply observed outcomes in both numerator and denominator, and these probabilities are fit by gradient ascent on the ratio. This is a direct fit to the observed outcome labels, not an unbiased estimator of a counterfactual quantity: the objective can increase by concentrating mass on high-outcome treated users regardless of their true causal response. The IPW form in Eq. (10) and Appendix B appears only as an evaluation variant; the main objective Eq. (9) does not include propensity terms, so the causal interpretation of SCPM's ranking is unsupported.
  3. [Section 3.3 and Algorithm 1] The recursive stacking architecture is justified by the Naive Bayes assumptions I_x independent of rho_c given x and, for additional factors, analogous conditional independences. These assumptions are stated without justification and are not checked empirically; if treatment assignment or intensity is confounded with the outcome, the learned posterior p(I_x | Pi_x, x) has no causal reading. The paper provides no sensitivity analysis or diagnostic, despite the fact that the entire ranking procedure rests on these independence conditions.
  4. [Section 4.3, Table 1] The claim that SCPM 'consistently performs on three out of four evaluation metrics' is not accurate: SCPM's KRCC (0.2098 plus/minus 0.0443) is lower than RERUM CFR (was) (0.2144 plus/minus 0.0165). In addition, no significance tests are reported, and the AUUC and AUQC gains over RERUM CFR (was) are within roughly 0.01, so the statement that SCPM 'significantly outperforms' the baselines is not established.
minor comments (4)
  1. [Section 3.4, first paragraph after Eq. (6)] The text contains typos: 'Te' should be 'The' and 'Wholistic' should be 'Holistic'.
  2. [Section 3.2, Eq. (2)] The summation index x is also used for the covariate vector, which makes the expression difficult to parse; please use a separate index such as i for cohort members.
  3. [Section 4.2, first paragraph] The sentence 'we intend to makes Ponpare dataset a valuable benchmark' contains a grammatical error and should be rewritten.
  4. [Appendix D, R-Learner with Multi-layer Perceptron] The text reports 'the optimal number of hidden layers to be 92' for US Census and '100' for Covertype; these values are almost certainly hidden units, not layers, and should be clarified.

Circularity Check

2 steps flagged · score 7.0 of 10

The central derivation of E[Yρc|x] in Eq. 2 defines the counterfactual expectation as a normalized weighted sum of observed treated outcomes, and the SCPM objective then optimizes exactly those weights against the same observed outcomes, so the 'prediction' reduces to the fitted input by construction.

  1. self definitional [Section 3.2, Eq. (2)]
    "We expand E[Yρc|x] from Eq. 1 in the case where ρc is observed, such as in a case where the training set has treatment and controlled cohorts: E[Yρc|x] = Σ_{x∈cohort,ρc≠0} p(Ix|ρc,x)Yρc. (2) Here, we consider the probability p(Ix|ρc,x) as the posterior probability of the subject being chosen for treatment with observed ρc and covariates x in the treatment cohort."

    The right-hand side is a convex combination of observed outcomes from the treated cohort only, with weights p(Ix|ρc,x) that are learned neural-network scores normalized by a partition function (Eq. 4). The equality is asserted, not derived from ignorability, overlap, or inverse propensity weighting, and no conditioning on a fixed dose ρc is actually enforced because the sum runs over subjects with different observed doses. Thus E[Yρc|x] is defined to be the weighted sum of observed outcomes rather than identified from causal assumptions; the subsequent 'prediction' is this same weighted sum by construction.

  2. fitted input called prediction [Section 3.4, Eqs. (7) and (9)]
    "τ(Πx,θ,x) = E(Y1−Y0|x,θ) = Σ_{Ti=1} p_i(Ix|Πx,θ,x)Y(i) − Σ_{Ti=0} p_i(Ix|Πx,θ,x)Y(i). (7) ... ˆθ = argmaxθ [τ∗r(Πx,θ,x)/τ∗c(Πx,θ,x) − Λn(·)]. (9)"

    The p_i in Eq. 7 are the NANBL normalized effectiveness scores, and Eq. 9 maximizes the ratio of two weighted sums of observed outcomes over the model parameters. The parameters are therefore fitted to make the weighted observed-outcome sum large, and the predicted treatment effect is exactly that fitted weighted sum. No propensity-weighted or otherwise identified counterfactual appears in the main objective; the paper's own IPW form (Eq. 10) is introduced only in Section 3.5 and used as a separate evaluation variant. The claimed prediction therefore reduces to the fitted input by construction.

full rationale

The main circularity is in the identification step, not in the benchmarking. Equation (2) is presented as a derivation 'from first principles' of E[Yρc|x], but it is actually a definition of the estimand as a normalized weighted average of observed treated outcomes. The NANBL weights p(Ix|ρc,x) are themselves learned by maximizing the same weighted outcome sum through Eq. 7 and Eq. 9, so the central 'prediction' is the fitted sum by construction. The paper's own Section 3.5 acknowledges that propensity weighting is needed when treatment is not random, yet the main SCPM objective omits it; the IPW form is relegated to Appendix B and Table 2. There is no load-bearing self-citation problem: the citation to the authors' prior direct-ranking work [14] is used as background, and the equations are re-derived in this paper. The external benchmark comparisons are real experiments, so the paper is not wholly tautological, but the validity of the estimator for the counterfactual quantity it claims to estimate is not established by the derivation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The model rests on a chain of assumptions: the Naive Bayes decomposition, the unproven equality in Eq. 2, unconfoundedness for the propensity extension, and Slater's condition for the dual optimization. The most damaging is the Eq. 2 equality, which is not stated as an assumption in the paper but is load-bearing.

free parameters (3)
  • Lambda (Lagrangian multiplier) = 0.1 for Ponpare, 0.05 for US Census and Covertype
    Used to combine incremental value and incremental cost into a single score in Duality R-Learner; selected by hyperparameter search on validation.
  • Barrier temperature T = starting 0.5, +0.1 every 10 Adam steps
    Controls the softness of the constraint barrier in Constrained Ranking; annealing schedule chosen by cross-validation.
  • Regularization Lambda = not specified
    Regularization term in the SCPM objective (Eq. 9); its form and value are not given.
assumptions (4)
  • ad hoc to paper Naive Bayes independence: Ix is independent of rho_c given x, and similarly for treatment assignment t_a.
    Made to factor the posterior into neural likelihoods; no justification on real data.
  • ad hoc to paper Eq. 2 equality: the normalized weighted sum over the treated cohort equals E[Y_{rho_c}|x].
    Assumed to connect the NANBL probabilities to the outcome expectation; not derived or justified.
  • domain assumption Unconfoundedness: E(T|X,Y1) = E(T|X).
    Standard causal identification assumption invoked in Appendix B for propensity weighting.
  • ad hoc to paper Slater's condition holds for the budget-constrained problem.
    Asserted in Appendix C to justify strong duality; the proof is hand-wavy ('we can always satisfy by selecting sufficiently small values').

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning of Continuous and Structured Policies for Aggregated Heterogeneous Treatment Effects." pith.science (2026). https://pith.science/paper/Y3NYDRXL

@misc{pith2026250705511,
  author       = {Pith},
  title        = {Pith review of: Deep Learning of Continuous and Structured Policies for Aggregated Heterogeneous Treatment Effects},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y3NYDRXL}},
  note         = {Machine review of arXiv:2507.05511}
}
read the original abstract

As estimation of Heterogeneous Treatment Effect (HTE) is increasingly adopted across a wide range of scientific and industrial applications, the treatment action space can naturally expand, from a binary treatment variable to a structured treatment policy. This policy may include several policy factors such as a continuous treatment intensity variable, or discrete treatment assignments. From first principles, we derive the formulation for incorporating multiple treatment policy variables into the functional forms of individual and average treatment effects. Building on this, we develop a methodology to directly rank subjects using aggregated HTE functions. In particular, we construct a Neural-Augmented Naive Bayes layer within a deep learning framework to incorporate an arbitrary number of factors that satisfies the Naive Bayes assumption. The factored layer is then applied with continuous treatment variables, treatment assignment, and direct ranking of aggregated treatment effect functions. Together, these algorithms build towards a generic framework for deep learning of heterogeneous treatment policies, and we show their power to improve performance with public datasets.

Figures

Figures reproduced from arXiv: 2507.05511 by the authors.

Figure 1
Figure 1. Bayesian Network across covariates, the continuous [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Recursively applying the NANBLs for an arbitrary [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Cost curve for US Census dataset [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Visualization of SCPM user embeddings for age [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

46 extracted references · 36 canonical work pages

  1. [1]

    [n. d.]. Grf: Generalized Random Forests. https://grf-labs.github.io/grf/. Accessed: 2019-11-15

    work page 2019

  2. [2]

    [n. d.]. US Census 1990 Dataset on UCI Machine Learning Repository. https: //archive.ics.uci.edu/ml/datasets/US+Census+Data+(1990). Accessed: 2019-11-15

    work page 1990

  3. [3]

    Meng Ai, Zhuo Chen, Jibin Wang, Jing Shang, Tao Tao, and Zhen Li. 2024. Im- prove ROI with Causal Learning and Conformal Prediction. In 2024 IEEE 40th International Conference on Data Engineering (ICDE) . IEEE, 598–610

    work page 2024

  4. [4]

    Michelle Andrews, Jody Goehring, Sam Hui, Joseph Pancras, and Lance Thornswood. 2016. Mobile Promotions: A Framework and Research Priorities. Journal of Interactive Marketing 34 (2016), 15–24. doi:10.1016/j.intmar.2016.03.004

    work page doi:10.1016/j.intmar.2016.03.004 2016

  5. [5]

    Susan Athey and Guido Imbens. 2016. Recursive partitioning for heterogeneous causal effects. Proceedings of the National Academy of Sciences 113, 27 (2016), 7353–7360

    work page 2016

  6. [6]

    Elie Azeraf, Emmanuel Monfrini, and Wojciech Pieczynski. 2021. Improving usual Naive Bayes classifier performances with Neural Naive Bayes based models. arXiv preprint arXiv:2111.07307 (2021)

    work page Pith review arXiv 2021

  7. [7]

    Dimitri P Bertsekas. 1999. Nonlinear programming. Athena scientific Belmont

    work page 1999

  8. [8]

    Artem Betlei, Eustache Diemert, and Massih-Reza Amini. 2021. Uplift Modeling with Generalization Guarantees(KDD ’21). Association for Computing Machinery, New York, NY, USA, 55–65. doi:10.1145/3447548.3467395

    arXiv 2021

Show all 46 references
  1. [9]

    Jock Blackard. 1998. Covertype. UCI Machine Learning Repository. DOI: https://doi.org/10.24432/C50K5N

    1998 doi

  2. [10]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901

    2020

  3. [11]

    Tianqi Chen and Carlos Guestrin. 2016. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining . ACM, 785–794

    2016

  4. [12]

    Victor Chernozhukov, Mert Demirer, Greg Lewis, and Vasilis Syrgkanis. 2019. Semi-parametric efficient policy learning with continuous actions. Advances in Neural Information Processing Systems 32 (2019)

    2019

  5. [13]

    Maxime Cohen, Michael-David Fiszer, and Baek Jung Kim. 2018. Frustration- Based Promotions: Field Experiments in Ride-Sharing. (2018)

    2018

  6. [14]

    Shuyang Du, James Lee, and Farzin Ghaffarizadeh. 2019. Improve user retention with causal learning. In The 2019 ACM SIGKDD Workshop on Causal Discovery . PMLR, 34–49

    2019

  7. [15]

    haisland0909, Shingo KATO, suharay, and Will Cukierski. 2015. Coupon Pur- chase Prediction. https://kaggle.com/competitions/coupon-purchase-prediction. Kaggle

    2015

  8. [16]

    Basil Halperin, Benjamin Ho, John List, Ian Muir, et al. 2018. Toward an under- standing of the economics of apologies: evidence from a large-scale natural field experiment. Technical Report. The Field Experiments Website

    2018

  9. [17]

    Hanna, Scott D

    Richard C. Hanna, Scott D. Swain, and Paul D. Berger. 2016. Optimizing time- limited price promotions. Journal of Marketing Analytics 4, 2 (July 2016), 77–92. doi:10.1057/s41270-016-0004-0

    2016 doi

  10. [18]

    Bowei He, Yunpeng Weng, Xing Tang, Ziqiang Cui, Zexu Sun, Liang Chen, Xiuqiang He, and Chen Ma. 2024. Rankability-enhanced Revenue Uplift Model- ing Framework for Online Marketing. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (Barcelo...

    2024

  11. [19]

    Po-Sen Huang, Xiaodong He, Jianfeng Gao, Li Deng, Alex Acero, and Larry Heck. 2013. Learning deep structured semantic models for web search using clickthrough data. In Proceedings of the 22nd ACM international conference on Information & Knowledge Management . 2333–2338

    2013

  12. [20]

    Kosuke Imai and David A Van Dyk. 2004. Causal inference with general treatment regimes: Generalizing the propensity score. J. Amer. Statist. Assoc. 99, 467 (2004), 854–866

    2004

  13. [21]

    Kevin Jarrett, Koray Kavukcuoglu, Marc’Aurelio Ranzato, and Yann LeCun. 2009. What is the best multi-stage architecture for object recognition?. In 2009 IEEE 12th international conference on computer vision . IEEE, 2146–2153

    2009

  14. [22]

    Nathan Kallus and Angela Zhou. 2018. Policy evaluation and optimization with continuous treatments. In International conference on artificial intelligence and statistics. PMLR, 1243–1251

    2018

  15. [23]

    Fahad Kamran, Maggie Makar, and Jenna Wiens. 2024. Learning to Rank for Opti- mal Treatment Allocation Under Resource Constraints. InProceedings of The 27th International Conference on Artificial Intelligence and Statistics (Proceedings of Ma- chine Learning Research, Vol. 238...

    2024

  16. [24]

    Noémi Kreif, Richard Grieve, Iván Díaz, and David Harrison. 2015. Evaluation of the effect of a continuous treatment: a machine learning approach with an application to treatment for traumatic brain injury. Health economics 24, 9 (2015), 1213–1228

    2015

  17. [25]

    Sören R Künzel, Jasjeet S Sekhon, Peter J Bickel, and Bin Yu. 2017. Meta-learners for Estimating Heterogeneous Treatment Effects using Machine Learning. arXiv preprint arXiv:1706.03461 (2017)

    2017 arXiv

  18. [26]

    Praveen Lalwani, Manas Kumar Mishra, Jasroop Singh Chadha, and Pratyush Sethi. 2022. Customer churn prediction system: a machine learning approach. Computing 104, 2 (2022), 271–294

    2022

  19. [27]

    Christos Louizos, Uri Shalit, Joris M Mooij, David Sontag, Richard Zemel, and Max Welling. 2017. Causal effect inference with deep latent-variable models. Advances in neural information processing systems 30 (2017)

    2017

  20. [28]

    Lunceford and M

    J.K. Lunceford and M. Davidian. 2004. Stratification and weighting via the propensity score in estimation of causal treatment effects: A comparative study. (2004)

    2004

  21. [29]

    Manzoor and Leman Akoglu

    Emaad A. Manzoor and Leman Akoglu. 2017. RUSH!: Targeted Time-limited Coupons via Purchase Forecasts. In KDD

    2017

  22. [30]

    X Nie and S Wager. 2017. Quasi-oracle estimation of heterogeneous treatment effects

    2017

  23. [31]

    Nocedal and S

    J. Nocedal and S. J. Wright. 2006. Numerical Optimization (2nd ed.) . Springer

    2006

  24. [32]

    Scott Powers, Junyang Qian, Kenneth Jung, Alejandro Schuler, Nigam H Shah, Trevor Hastie, and Robert Tibshirani. 2017. Some methods for heterogeneous treatment effect estimation in high-dimensions. arXiv preprint arXiv:1707.00102 (2017)

    2017 arXiv

  25. [33]

    ROSENBAUM and DONALD B

    PAUL R. ROSENBAUM and DONALD B. RUBIN. 1983. The cen- tral role of the propensity score in observational studies for causal ef- fects. Biometrika 70, 1 (04 1983), 41–55. doi:10.1093/biomet/70.1.41 arXiv:https://academic.oup.com/biomet/article-pdf/70/1/41/662954/70-1-41.pdf

    1983 doi

  26. [34]

    Donald B Rubin. 1974. Estimating causal effects of treatments in randomized and nonrandomized studies. Journal of educational Psychology 66, 5 (1974), 688

    1974

  27. [35]

    Piotr Rzepakowski and Szymon Jaroszewicz. 2012. Decision trees for uplift mod- eling with single and multiple treatments. Knowledge and Information Systems 32, 2 (2012), 303–327

    2012

  28. [36]

    Johansson, and David Sontag

    Uri Shalit, Fredrik D. Johansson, and David Sontag. 2017. Estimat- ing individual treatment effect: generalization bounds and algorithms. arXiv:1606.03976 [stat.ML] https://arxiv.org/abs/1606.03976

    2017 arXiv

  29. [37]

    Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, and Jeff Dean. 2017. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer. arXiv preprint arXiv:1701.06538 (2017)

    2017 arXiv

  30. [38]

    Claudia Shi, David Blei, and Victor Veitch. 2019. Adapting Neural Networks for the Estimation of Treatment Effects. InAdvances in Neural Information Processing Systems, H. Wallach, H. Larochelle, A. Beygelzimer, F. d'Alché-Buc, E. Fox, and R. Garnett (Eds.), Vol. 32. Curran As...

    2019

  31. [39]

    Zexu Sun, Hao Yang, Dugang Liu, Yunpeng Weng, Xing Tang, and Xiuqiang He. 2024. End-to-End Cost-Effective Incentive Recommendation under Budget Constraint with Uplift Modeling. arXiv:2408.11623 [cs.IR] https://arxiv.org/abs/ 2408.11623

    2024 arXiv

  32. [40]

    Irfan Ullah, Basit Raza, Ahmad Kamran Malik, Muhammad Imran, Saif Ul Islam, and Sung Won Kim. 2019. A churn prediction model using random forest: analysis of machine learning techniques for churn prediction and factor identification in telecom sector. IEEE access 7 (2019), 601...

    2019

  33. [41]

    Vafeiadis, K.I

    T. Vafeiadis, K.I. Diamantaras, G. Sarigiannidis, and K.Ch. Chatzisavvas. 2015. A comparison of machine learning techniques for customer churn prediction. Simulation Modelling Practice and Theory 55 (2015), 1–9. doi:10.1016/j.simpat. 2015.03.003

    2015 doi

  34. [42]

    Hamed Valizadegan, Rong Jin, Ruofei Zhang, and Jianchang Mao. 2009. Learning to rank by optimizing ndcg measure. Advances in neural information processing systems 22 (2009)

    2009

  35. [43]

    Laurens van der Maaten and Geoffrey Hinton. 2008. Visualizing Data using t-SNE. Journal of Machine Learning Research 9, 86 (2008), 2579–2605. http: //jmlr.org/papers/v9/vandermaaten08a.html

    2008

  36. [44]

    Stefan Wager and Susan Athey. 2017. Estimation and inference of heterogeneous treatment effects using random forests. J. Amer. Statist. Assoc. just-accepted (2017)

    2017

  37. [45]

    Stefan Wager and Susan Athey. 2018. Estimation and inference of heterogeneous treatment effects using random forests. J. Amer. Statist. Assoc. 113, 523 (2018), 1228–1242

    2018

  38. [46]

    Hao Zhou, Shaoming Li, Guibin Jiang, Jiaqi Zheng, and Dong Wang. 2023. Di- rect heterogeneous causal learning for resource allocation problems in market- ing. In Proceedings of the Thirty-Seventh AAAI Conference on Artificial Intelli- gence and Thirty-Fifth Conference on Innov...

    2023 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.