Pith. sign in

REVIEW 2 major objections 6 minor 41 references

Semiparametric Expectile Regression for High-dimensional Heavy-tailed and Heterogeneous Data

T0 review · 2 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A penalized partially linear additive expectile regression with SCAD or MCP penalty recovers the sparse oracle fit under heavy-tailed errors; the allowable covariate dimension grows as a power of n set by the finite moment k of the error.

desk verdict A clean extension of expectile regression to partially linear additive models with heavy tails, but the oracle theorem as printed rests on a missing minimum-eigenvalue condition. read the letter →

arxiv 1908.06431 v1 pith:3R42RGJC submitted 2019-08-18 math.ST stat.MEstat.TH

Jun Zhao , Guan'ao Yan , Yi Zhang This is my paper · ORCID
classification math.STstat.MEstat.TH MSC 62G0862J0762G20
keywords expectileregressionpartiallylinearadditivemodelheavy-tailederrorsheterogeneityoraclepropertySCADMCPB-splineapproximation
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 establish that regularized expectile regression works in a semiparametric high-dimensional setting even when regression errors are heavy-tailed with only finitely many moments. It proposes a partially linear additive model whose linear coefficients are penalized with a folded-concave penalty (SCAD or MCP) and whose additive nonlinear functions are fitted by B-splines. The main theorem says that, with probability tending to one, the oracle estimator—the fit one would get knowing the active variables—is a local minimum of the penalized objective, and the number $p$ of linear covariates can grow as a power of the sample size set by the error's finite moment $k$. This matters because heterogeneity and heavy tails are common in genetics and finance, the expectile loss is differentiable and avoids quantile-crossing problems, and sweeping the expectile level $\alpha$ reveals how covariates affect different parts of the conditional distribution.

What carries the argument

The central object is the asymmetric squared loss $\phi_\alpha(r) = |\alpha - I(r<0)|r^2$, whose minimizer defines the $\alpha$-expectile. Three pieces carry the argument: first, the B-spline approximation of each additive function $g_{0j}$, with the linear part separated by a weighted projection of the active covariates $x_A$ onto the spline basis using weights $w_i = E[\phi_\alpha(\epsilon_i)|x_i,z_i]$; second, the difference-of-convex decomposition of the SCAD/MCP penalized loss, which turns the search for local minima into a subdifferential intersection condition from convex analysis; and third, concentration bounds—Bernstein's inequality for the local quadratic terms and a moment inequality for the noise—that convert the finite-$k$ moment assumption into the dimension allowance $p = o((n\lambda^2)^k)$.

What would settle it

A direct check would be a simulation with an active linear covariate equal to a spline function of $z$ plus a tiny independent perturbation, so that the leftover design matrix is nearly singular while Conditions 3.2–3.5 still hold; if the oracle estimator ceases to be a local minimizer there, the theorem as stated fails. A second check uses t-distributed errors with only about two finite moments and $p = n^{0.6}$: the paper's bound $O(p(n\lambda^2)^{-k})$ then no longer vanishes, so the empirical frequency of the oracle local-minimum event should visibly drop.

Watch

Extended reading notes

Core claim

The paper's core claim is Theorem 3.2: under Conditions 3.1–3.5, if the tuning parameter satisfies $\lambda = o(n^{-(1-C_4)/2})$, with $q_n = o(n\lambda^2)$, $k_n = o(n\lambda^2)$, and $p = o((n\lambda^2)^k)$, then with probability tending to one the oracle estimator $(\hat{\beta}^*, \hat{\xi}^*)$ is a local minimum of the penalized expectile loss. This implies that the sparse linear coefficients and the additive nonlinear functions are simultaneously recoverable at a fixed expectile level $\alpha$ even though the errors have only $2k$ finite moments and the error distribution may be heteroscedastic. The dimension constraint $p = o(n^{C_4 k})$ makes the moment condition the limiting factor: heavier tails cap the dimensionality at a lower power of $n$, while sub-Gaussian errors allow $p$ to grow as any polynomial in $n$.

Load-bearing premise

The proof needs the leftover parts of the active linear covariates, after subtracting their spline-fitted nonlinear pieces, to remain well spread out in all directions; the paper's Condition 3.2 only bounds how large these designs can be, not how small their spread can get, yet Lemmas 6.9 and 6.12 rely on that small-spread bound.

Editorial extensions

If this is right

  • At expectile level $\alpha$, the oracle estimator is consistent for the sparse linear coefficients at rate $O_p(\sqrt{q_n/n})$ and for the additive functions at $L_2$ rate $O_p(n^{-1}(q_n+k_n))$ when the theorem's conditions hold.
  • The dimension of the linear part may grow as $p = o(n^{C_4 k})$; every additional finite error moment buys a higher allowable power of $n$, and sub-Gaussian errors allow $p$ to be any polynomial in $n$.
  • Variables that affect only the conditional scale are invisible at $\alpha = 0.5$ but are selected at asymmetric levels such as $\alpha = 0.1$ and $0.9$, so expectile sweeping can detect heteroscedasticity.
  • The two-step LLA algorithm with SCAD/MCP approximates the nonconvex problem by convex subproblems; in simulations it estimates and selects more accurately than the Lasso version E-Lasso.
  • On the birth-weight gene expression data, different expectile levels select different gene sets while gestational age is selected at all levels, an indication of heterogeneity in the data.

Reading between the lines

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

  • A testable extension is to replace the bounded-eigenvalue condition on the projected design with a restricted eigenvalue condition; this could extend the oracle property to designs where active linear covariates are nearly collinear with the spline space.
  • The moment-dependence of $p$ suggests a practical diagnostic: estimate the residual tail index and then choose $\lambda$ and the nominal dimension cap as $n^{C_4 k}$; the method's reliability should degrade sharply when $k$ is only slightly above 1.
  • Because the weighted projection weights $w_i$ encode the expectile level, the same proof template should carry over to other asymmetric convex losses with quadratic curvature bounds, such as an asymmetric Huber loss smoothed at zero.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 6 minor

Summary. The paper develops a penalized partially linear additive expectile regression procedure for high-dimensional data with heavy-tailed heterogeneous errors. The nonparametric components are approximated by B-splines and the linear coefficients are penalized by a folded concave penalty (SCAD or MCP). The main theoretical claim is Theorem 3.2: under Conditions 3.1–3.5, the oracle estimator is, with probability tending to one, a local minimizer of the nonconvex empirical loss, with the allowable dimension p growing like a power of n determined by the number of finite error moments. The paper also proposes a two-step LLA-based algorithm, studies finite-sample behavior in simulations, and applies the method to a birth-weight gene-expression data set.

Significance. If the main theorem is correct, the paper extends expectile regression to a sparse semiparametric high-dimensional setting under moment conditions rather than sub-Gaussian tail assumptions, and it makes a concrete point: the growth rate of p is tied to the kth moment of the error. The proof strategy is conventional and extensive, using the strong convexity of the asymmetric quadratic loss, Bernstein and moment inequalities, spline approximation bounds, and the DC-programming characterization of local minima. The simulation comparison between SCAD and Lasso penalties and the heterogeneity-detection illustration are useful. The main caveat is that the stated assumptions do not currently support a key step in the proof, so the central result is not yet verified as printed.

major comments (2)
  1. [Condition 3.2 (Section 3.1) and Lemma 6.9 (Appendix 6.3)] Condition 3.2 as printed bounds only the largest eigenvalues of n^{-1}X_AX_A' and n^{-1}Δ_nΔ_n' above and below. The proof of Lemma 6.9, however, needs W_n = n^{-1}∑_{i=1}^n w_i δ_iδ_i' to be positive definite with λ_min(W_n) bounded away from zero, and Lemma 6.6 requires W_n^{-1} to exist. A lower bound on λ_max does not control λ_min: if the vectors δ_i lie in a proper subspace of R^{q_n}, one can have λ_max(n^{-1}Δ_nΔ_n') ≥ C_1 while λ_min(W_n)=0, in which case the quadratic form (θ_1−tildeθ_1)'W_n(θ_1−tildeθ_1) vanishes along a nonzero direction and the key display (6.4) does not follow. This gap is load-bearing because Lemma 6.9 feeds both Theorem 3.1 and Theorem 3.2. The repair is standard: add a lower bound such as λ_min(n^{-1}Δ_n'Δ_n) ≥ C_1, or a restricted-eigenvalue condition on the weighted projected design, and then verify Lemmas 6.6 and 6.9 under the corrected condition.
  2. [Theorem 3.2 and Section 3.2 scaling assumptions] The theorem statement should make explicit that nλ^2 is required to diverge in order for the high-dimensional interpretation p = o((nλ^2)^k) to be non-vacuous. As written, λ = o(n^{-(1−C_4)/2}) alone permits values with nλ^2 → 0, in which case q_n = o(nλ^2), k_n = o(nλ^2), and p = o((nλ^2)^k) are compatible only with bounded or vanishing model dimensions. Since the paper's stated novelty is the power-of-n growth of p allowed by the moment condition, the authors should state explicitly that p, q_n, and k_n diverge, or at least that nλ^2 → ∞, and adjust the sentence following (3.8) accordingly.
minor comments (6)
  1. [Appendix 6.2 notation] The matrix W_n is defined as n^{-1}∑ w_i δ_iδ_i' and simultaneously declared to be in R^{n×n}; since δ_i ∈ R^{q_n}, this matrix is q_n × q_n, and Lemma 6.3(2) uses it as an approximation to n^{-1}X^{*\prime}B_nX^* ∈ R^{q_n×q_n}. Please correct the displayed dimension.
  2. [Appendix 6.2 notation (W_B)] The symbol W_B is used both as W_{2B} = W'B_nW and, through W_B^{-1} in \tilde W(z_i) and θ_2, as if it were a square root of W'B_nW. The proof of Lemma 6.2(3) also switches between ||W_B^{-1}|| and λ_min(W'B_nW)^{-1/2}. Please define W_B explicitly, for example as (W'B_nW)^{1/2}, and make the norm identities consistent.
  3. [Section 2.2, formula for H_λ] In the display after equation (2.8), the term [λ|θ| − (a+1)^2/2] I(|θ| > aλ) should presumably read [λ|θ| − (a+1)λ^2/2] I(|θ| > aλ); the missing λ^2 makes the stated SCAD decomposition dimensionally inconsistent.
  4. [Section 5 and Table 3] The prediction errors L_1 and L_2 are defined with denominator 1/24∑_{i∈test set}, but the test set in the random-partition procedure has size 15; the 'All Data' rows of Table 3 report L_1 and L_2 without a test set, so those numbers cannot be reproduced from the stated procedure. Please reconcile the formulas, the sample sizes, and the table entries.
  5. [Abstract and keywords] There are several typos: 'Monto Carlo' should be 'Monte Carlo' and 'Addtive' should be 'Additive'. These do not affect the scientific content but should be corrected.
  6. [Section 5 introductory paragraph] The text cites subsamples of sizes n=20 and n=52 from Votavova et al. (2011) but then says the data set has 65 observations; please clarify whether 65 refers to the subset used here or is a typo, and state how the clinical variables and gene-expression measurements are matched.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the oracle local-minimizer claim is proved from stated conditions; the only serious issue is a missing minimum-eigenvalue condition in Condition 3.2, which is a correctness gap, not a circular step.

full rationale

There is no circular step in the claimed derivation chain. The oracle estimator is defined as the unpenalized minimizer over the true active set in (3.1), and Theorem 3.2 proves, rather than assumes, that this estimator is a local minimizer of the penalized objective (2.7)-(2.8) by verifying the subgradient intersection condition of Lemma 6.10 using the score bounds in Lemma 6.12. The beta-min condition (3.5) and the rate constraints lambda = o(n^{-(1-C4)/2}), q_n = o(n lambda^2), k_n = o(n lambda^2), p = o((n lambda^2)^k) are hypotheses, and the dimension bound p = o(n^{C4 k}) is a derived conclusion obtained from Markov's inequality in Lemma 6.12, not an input. The only citation to the authors' own prior work, Zhao et al. (2018), appears in the introduction and in the remark after Condition 3.1 to motivate the finite-moment heavy-tailed error assumption; it is not used to prove Theorem 3.2 or any supporting lemma, so it is not load-bearing. The proof's main weakness is a different issue: Condition 3.2 as printed bounds only lambda_max of n^{-1}X_A X_A' and n^{-1}Delta_n Delta_n', while the proof of Lemma 6.9 and the use of W_n^{-1} in Lemma 6.6 require W_n = n^{-1} sum_i w_i delta_i delta_i' to be positive definite. The proof states 'from Condition 3.2, for any theta_1 satisfying ||theta_1 - tilde theta_1|| >= M > 0, we have 1/(2n)(theta_1 - tilde theta_1)' W_n (theta_1 - tilde theta_1) > 0,' but the printed condition does not imply this. This is a missing-assumption correctness gap, not a circularity: adding a minimum-eigenvalue or restricted-eigenvalue condition would be a genuine additional input, not a restatement of the theorem. The theorem, the lemmas, and the simulations are otherwise self-contained against external benchmarks, and no fitted parameter is renamed as a prediction.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The main inputs are moment, design, sparsity, beta-min, and spline smoothness assumptions. No new particles, forces, dimensions, or latent quantities are introduced. The oracle estimator is a benchmark, not an invented entity. The as-written Condition 3.2 is a gap because the proof requires a minimum eigenvalue condition that is not stated.

free parameters (3)
  • tuning parameter lambda = chosen by tuning set in simulations and 5-fold CV in the real data; asymptotic rate lambda = o(n^{-(1-C4)/2})
    Controls sparsity in the SCAD/MCP penalty; Theorem 3.2 only holds for lambda in a shrinking window, and all finite-sample performance depends on its choice.
  • SCAD shape parameter a = 3.7
    Set following Fan and Li (2001); not fitted to data but a user choice that affects finite-sample selection and is fixed in all simulations.
  • B-spline basis count k_n = 3 cubic B-spline basis functions per component in simulations; theory sets k_n approximately n^{1/(2r+1)}
    Smoother complexity is chosen by the user; approximation rates and the theorem rates depend on k_n.
assumptions (6)
  • domain assumption Conditional expectile identifies the model: m_alpha(epsilon_i | x_i, z_i) = 0, so beta* and g0 minimize the population expectile risk.
    Section 2.1, after model (2.1). If the conditional expectile of the error is not zero, the intercept and nonparametric part absorb a shift.
  • domain assumption Finite 2k-th moment of errors, E(epsilon_i^{2k} | x_i, z_i) < C, with k >= 1.
    Condition 3.1 replaces the sub-Gaussian assumption and controls all stochastic bounds in the Appendix.
  • domain assumption Design conditions: bounded x_ij and eigenvalue control of X_A and Delta_n.
    Condition 3.2 is printed with lambda_max bounds only, but the proof of Lemma 6.9 needs lambda_min(W_n) > 0. This is the weakest printed assumption.
  • domain assumption Sparsity and beta-min: q_n = O(n^{C3}) with C3 < 1/2, and n^{(1-C4)/2} min |beta*_j| >= C5.
    Conditions 3.4 and 3.5 are needed for selection consistency and to separate active from inactive coefficients.
  • standard math B-spline approximation results for g0 in H^r with r > 1.5.
    Used in Lemma 6.2, Lemma 6.5, and in bounds on the approximation error u_ni, citing Stone (1985) and Schumaker (2007).
  • standard math DC programming local minimizer criterion of Tao and An (1997).
    Lemma 6.10 is the basis for proving that the oracle estimator is a local minimum of the k minus l decomposition.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Semiparametric Expectile Regression for High-dimensional Heavy-tailed and Heterogeneous Data." pith.science (2026). https://pith.science/paper/3R42RGJC

@misc{pith2026190806431,
  author       = {Pith},
  title        = {Pith review of: Semiparametric Expectile Regression for High-dimensional Heavy-tailed and Heterogeneous Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3R42RGJC}},
  note         = {Machine review of arXiv:1908.06431}
}
abstract

Recently, high-dimensional heterogeneous data have attracted a lot of attention and discussion. Under heterogeneity, semiparametric regression is a popular choice to model data in statistics. In this paper, we take advantages of expectile regression in computation and analysis of heterogeneity, and propose the regularized partially linear additive expectile regression with nonconvex penalty, for example, SCAD or MCP for such high-dimensional heterogeneous data. We focus on a more realistic scenario: the regression error is heavy-tailed distributed and only has finite moments, which is violated with the classical sub-gaussian distribution assumption and more common in practise. Under some regular conditions, we show that with probability tending to one, the oracle estimator is one of the local minima of our optimization problem. The theoretical study indicates that the dimension cardinality of linear covariates our procedure can handle with is essentially restricted by the moment condition of the regression error. For computation, since the corresponding optimization problem is nonconvex and nonsmooth, we derive a two-step algorithm to solve this problem. Finally, we demonstrate that the proposed method enjoys good performances in estimation accuracy and model selection through Monto Carlo simulation studies and a real data example. What's more, by taking different expectile weights $\alpha$, we are able to detect heterogeneity and explore the entire conditional distribution of the response variable, which indicates the usefulness of our proposed method for analyzing high dimensional heterogeneous data.

Figures

Figures reproduced from arXiv: 1908.06431 by the authors.

Figure 1
Figure 1. Boxplots of Prediction Errors [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 40 canonical work pages

  1. [1]

    J., Amemiya, T., and Poirier, D

    Aigner, D. J., Amemiya, T., and Poirier, D. J. (1976). On the estimation of production frontiers: maximum likelihood estimation of the parameters of a discontinuous density function. International Economic Review, 17(2), 377-396

    work page 1976

  2. [2]

    Aitkin, M. (1987). Modelling variance heterogeneity in normal regression using GLIM. Journal of the Royal Statistical Society: Series C (Applied Statistics), 36(3), 332-339

    work page 1987

  3. [3]

    B\" u hlmann, P., and Van De Geer, S. (2011). Statistics for high-dimensional data: methods, theory and applications. Springer Science & Business Media

    work page 2011

  4. [4]

    and Zhao, L

    Buja, A., Berk, R., Brown, L., George, E., Pitkin, E., Traskin, M., ... and Zhao, L. (2014). Models as approximations, part i: A conspiracy of nonlinearity and random regressors in linear regression. arXiv preprint arXiv:1404.1578

    arXiv 2014

  5. [5]

    Chen, H. (1988). Convergence rates for parametric components in a partly linear model. The Annals of Statistics, 16(1), 136-146

    work page 1988

  6. [6]

    Chung, K. (2008). The strong law of large numbers. Selected Works of Kai Lai Chung, 145-156

    work page 2008

  7. [7]

    J., Chen, J., and Li, H

    Daye, Z. J., Chen, J., and Li, H. (2012). High - Dimensional Heteroscedastic Regression with an Application to eQTL Data Analysis. Biometrics, 68(1), 316-326

    work page 2012

  8. [8]

    G., and Newey, W

    Donald, S. G., and Newey, W. K. (1994). Series estimation of semilinear models. Journal of Multivariate Analysis, 50(1), 30-40

    work page 1994

Show all 41 references
  1. [9]

    Fan, J., Li, Q., and Wang, Y. (2017). Estimation of high dimensional mean regression in the absence of symmetry and light tail assumptions. Journal of the Royal Statistical Society, 79(1), 247-265

    2017

  2. [10]

    Fan, J., and Li, R. (2001). Variable selection via nonconcave penalized likelihood and its oracle properties. Journal of the American statistical Association, 96(456), 1348-1360

    2001

  3. [11]

    Fan, J., and Lv, J. (2008). Sure independence screening for ultrahigh dimensional feature space. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 70(5), 849-911

    2008

  4. [12]

    Fan, J., Xue, L., and Zou, H. (2014). Strong oracle optimality of folded concave penalized estimation. The Annals of statistics, 42(3), 819

    2014

  5. [13]

    Gu, Y., and Zou, H. (2016). High-dimensional generalizations of asymmetric least squares regression and their applications. The Annals of Statistics, 44(6), 2661-2694

    2016

  6. [14]

    Gy\" o rfi, L., Kohler, M., Krzyzak, A., and Walk, H. (2006). A distribution-free theory of nonparametric regression. Springer Science & Business Media

    2006

  7. [15]

    Hastie, T. J. and Tibshirani, R. J. (1990). Generalized Additive Models. Chapman and Hall, London

    1990

  8. [16]

    He, X., Wang, L., and Hong, H.G. (2013). Quantile-adaptive model-free variable screening for high-dimensional heterogeneous data. The Annals of Statistics, 41(1), 342-369

    2013

  9. [17]

    Horowitz, J. L. (1999). Semiparametric estimation of a proportional hazard model with unobserved heterogeneity. Econometrica, 67(5), 1001-1028

    1999

  10. [18]

    Kim, Y., Choi, H., and Oh, H. S. (2008). Smoothly clipped absolute deviation on high dimensions. Journal of the American Statistical Association, 103(484), 1665-1673

    2008

  11. [19]

    Kramer, M. S. (1987). Determinants of low birth weight: methodological assessment and meta-analysis. Bulletin of the World Health Organization, 65(5), 663-737

    1987

  12. [20]

    Kudo, C., Ajioka, I., Hirata, Y., and Nakajima, K. (2005). Expression profiles of EphA3 at both the RNA and protein level in the developing mammalian forebrain. Journal of Comparative Neurology, 487(3), 255-269

    2005

  13. [21]

    Liang, H., and Li, R. (2009). Variable selection for partially linear models with measurement errors. Journal of the American Statistical Association, 104(485), 234-248

    2009

  14. [22]

    Y., Wang J., Huang F., et al

    Lv X. Y., Wang J., Huang F., et al. (2018). ppEphA3 contributes to tumor growth and angiogenesis in human gastric cancer cells. Oncology Reports, 40(4), 2408-2416

    2018

  15. [23]

    Michael, G., and Stephen, B. (2013). CVX: Matlab software for disciplined convex programming, version 2.0 beta. http://cvxr.com/cvx

    2013

  16. [24]

    Michael, G., and Stephen, B. (2008). Graph implementations for nonsmooth convex programs, Recent Advances in Learning and Control (a tribute to M. Vidyasagar), V. Blondel, S. Boyd, and H. Kimura, editors, pages 95-110, Lecture Notes in Control and Information Sciences, Springe...

    2008

  17. [25]

    National Research Council. (2013). Frontiers in massive data analysis. National Academies Press

    2013

  18. [26]

    K., and Powell, J

    Newey, W. K., and Powell, J. L. (1987). Asymmetric least squares estimation and testing. Econometrica: Journal of the Econometric Society, 55(4), 819-847

    1987

  19. [27]

    A., and Stasinopoulos, D

    Rigby, R. A., and Stasinopoulos, D. M. (1996). A semi-parametric additive model for variance heterogeneity. Statistics and Computing, 6(1), 57-65

    1996

  20. [28]

    Robinson, P. M. (1988). Root-N-consistent semiparametric regression. Econometrica, 56(4), 931-954

    1988

  21. [29]

    Schumaker, L. (2007). Spline functions: basic theory. Cambridge University Press

    2007

  22. [30]

    Sherwood, B., and Wang, L. (2016). Partially linear additive quantile regression in ultra-high dimension. The Annals of Statistics, 44(1), 288-317

    2016

  23. [31]

    Stone, C. J. (1985). Additive regression and other nonparametric models. The Annals of Statistics, 13(2), 689-705

    1985

  24. [32]

    D., and An, L

    Tao, P. D., and An, L. T. H. (1997). Convex analysis approach to dc programming: Theory, algorithms and applications. Acta Mathematica Vietnamica, 22(1), 289-355

    1997

  25. [33]

    Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society. Series B (Methodological), 58(1), 267-288

    1996

  26. [34]

    Turan, N., Ghalwash, M.F., Katari, S., Coutifaris, C., Obradovic, Z., Sapienza, C. (2012). DNA methylation differences at growth related genes correlate with birth weight: a molecular signature linked to developmental origins of adult disease? BMC Medical Genomics, 5(1), 10

    2012

  27. [35]

    W., and Wellner, J

    van der Vaart, A. W., and Wellner, J. A. (1996). Weak Convergence and Empirical Processes: with applications to statistics. New York: Springer, 552

    1996

  28. [36]

    D., Fejglova, K., Vasikova, A., Krejcik, Z., Pastorkova, A.,

    Votavova, H., Merkerova, M. D., Fejglova, K., Vasikova, A., Krejcik, Z., Pastorkova, A., ... & Brdicka, R. (2011). Transcriptome alterations in maternal and fetal cells induced by tobacco smoke. Placenta, 32(10), 763-770

    2011

  29. [37]

    S., Sobotka, F., Kneib, T., and Kauermann, G

    Waltrup, L. S., Sobotka, F., Kneib, T., and Kauermann, G. (2015). Expectile and quantile regressionDavid and Goliath? Statistical Modelling, 15(5), 433-456

    2015

  30. [38]

    Wang, L., Wu, Y., and Li, R. (2012). Quantile regression for analyzing heterogeneity in ultra-high dimension. Journal of the American Statistical Association, 107(497), 214-222

    2012

  31. [39]

    and Zhang Y

    Zhao J., Chen Y. and Zhang Y. (2018). Expectile regression for analyzing heteroscedasticity in high dimension. Statistics & Probability Letters, 137, 304-311

    2018

  32. [40]

    Zhang, C. H. (2010). Nearly unbiased variable selection under minimax concave penalty. The Annals of statistics, 38(2), 894-942

    2010

  33. [41]

    Zou, H., and Li, R. (2008). One-step sparse estimates in nonconcave penalized likelihood models. The Annals of statistics, 36(4), 1509-1533

    2008

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.