pith. sign in

arxiv: 2606.26136 · v1 · pith:QLEDS35Dnew · submitted 2026-06-18 · 💻 cs.SI · math.ST· stat.ME· stat.TH

Bayesian Predictive Synthesis for Dynamic Networks: Forecasting and Identifying Structural Mechanisms

Marios Papamichalis , Regina Ruane , Theofanis Papamichalis This is my paper

Pith reviewed 2026-06-26 14:44 UTC · model grok-4.3

classification 💻 cs.SI math.STstat.MEstat.TH
keywords dynamic networksBayesian predictive synthesisstructural mechanismsnetwork forecastingmechanism identificationlink predictiontime-varying weightsgraph snapshots
0
0 comments X

The pith

Bayesian predictive synthesis combines mechanism forecasts with time-varying weights to identify the dominant structure from one network snapshot.

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

The paper develops a synthesis approach in which each structural mechanism functions as an independent forecasting agent that predicts the next network edges, while a higher layer learns time-varying weights on those agents. This produces both a calibrated forecast of the next graph and an estimate of which mechanism is currently most informative. Under the stated identification theory, a single observed snapshot suffices to estimate the weights, a sharp threshold determines when mechanisms can be told apart, and switches between mechanisms are tracked at a per-switch cost that cannot be improved. For a static network the procedure collapses to calibrated link prediction. A sympathetic reader would care because many real networks are shaped by competing structures whose relative influence changes over time, yet most models commit to one fixed mechanism in advance.

Core claim

Dynamic Bayesian predictive synthesis represents each candidate mechanism as an agent that issues an edge forecast for the next time step; a synthesis layer then forms a convex combination of those forecasts using time-varying weights that are identifiable from observed snapshots under a sparse-safe parametrization. The resulting procedure returns calibrated edge probabilities together with inference on the active mechanism, separates distinguishable from indistinguishable mechanisms by a sharp threshold, tracks mechanism changes at optimal per-switch cost, and reduces exactly to calibrated link prediction when only one snapshot is available.

What carries the argument

The synthesis layer that forms time-varying convex combinations of mechanism-specific edge forecasts, together with the identification theory that recovers those weights from a single graph snapshot.

If this is right

  • The method yields both edge forecasts and statements about which mechanism is currently dominant, with valid intervals conditional on the fitted agents.
  • A change in the dominant mechanism is detected at the lowest possible per-switch cost permitted by the information in the snapshots.
  • When only one snapshot is observed the procedure supplies calibrated probabilities for each possible edge.
  • Mechanisms that fall below the sharp distinguishability threshold cannot have their weights reliably recovered.
  • The same synthesis layer can be applied to any collection of forecasting agents that produce edge probabilities.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The framework could be used to test whether a proposed new mechanism adds predictive value beyond an existing set of agents without refitting the entire model.
  • Because weights are recovered from a single snapshot, the approach may allow retrospective analysis of historical networks where only isolated observations survive.
  • If the independence assumption among agents is relaxed, the identification theory would need to be extended to account for correlated forecast errors.

Load-bearing premise

Mechanisms can be treated as independent forecasting agents whose weights remain identifiable from graph snapshots under the sparse-safe parametrization and identification conditions given in the paper.

What would settle it

Simulated dynamic networks in which the true active mechanism is known at every step, yet the estimated weights fail to concentrate on that mechanism or the method produces miscalibrated forecasts after a switch, would falsify the central claim.

Figures

Figures reproduced from arXiv: 2606.26136 by Marios Papamichalis, Regina Ruane, Theofanis Papamichalis.

Figure 1
Figure 1. Figure 1: S&P 500 correlation networks, 2013–2018. (Top) One-step-ahead predictive log-score; [PITH_FULL_IMAGE:figures/full_fig_p018_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Sequential one-step-ahead forecast performance on the S&P network, in the form used to [PITH_FULL_IMAGE:figures/full_fig_p019_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Pooling helps estimation where the snapshot is sparse. On a stable three-block model [PITH_FULL_IMAGE:figures/full_fig_p020_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: SocioPatterns high-school contact network, five school days (Theorem [PITH_FULL_IMAGE:figures/full_fig_p022_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: SocioPatterns hospital-ward contact network, four days (Theorem [PITH_FULL_IMAGE:figures/full_fig_p023_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Well-separated three-regime localization ( [PITH_FULL_IMAGE:figures/full_fig_p024_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The single-snapshot theory confirmed on the same data. (Left) Theorem [PITH_FULL_IMAGE:figures/full_fig_p025_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Two large dynamic networks (arXiv HEP-PH, Enron), agents fit on the discounted [PITH_FULL_IMAGE:figures/full_fig_p029_8.png] view at source ↗
read the original abstract

Networks are shaped by competing structural mechanisms, such as communities, geometry, or hubs. In a dynamic network the most predictive mechanism can change, and a model tied to one mechanism, or to fixed weights, cannot adapt as the dominant structure shifts. We develop dynamic Bayesian predictive synthesis for networks, in which a mechanism is an agent forecasting the next snapshot's edges and a synthesis layer combines them with time-varying weights. At each step the method returns a calibrated edge forecast and inference on the mechanism weights, with intervals valid given the fitted agents, so it also reports which mechanism is most informative. Inference of this kind requires a sparse-safe parametrization and an identification theory, under which a single graph identifies and estimates the weights. A sharp threshold separates distinguishable from indistinguishable mechanisms, a change in the active mechanism is tracked at an optimal per-switch cost, and for a single snapshot the method reduces to calibrated link prediction. On real networks, simulations, and benchmarks, the synthesis gives accurate, calibrated forecasts and recovers the leading mechanism when

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript develops dynamic Bayesian predictive synthesis for networks (BPSDN), modeling structural mechanisms (e.g., communities, geometry, hubs) as independent forecasting agents whose edge predictions are combined via time-varying weights in a synthesis layer. It supplies an identification theory under a sparse-safe parametrization under which a single observed graph snapshot identifies and estimates the weights, with a sharp threshold separating distinguishable from indistinguishable mechanisms, optimal per-switch tracking of mechanism changes, and reduction to calibrated link prediction in the static case. The method outputs calibrated forecasts together with valid intervals and mechanism-inference diagnostics. Empirical evaluation on real networks, simulations, and benchmarks is claimed to show accurate calibrated forecasts and recovery of the leading mechanism.

Significance. If the identification theory is internally consistent and the empirical calibration results hold under the stated assumptions, the framework would offer a principled way to adaptively forecast and diagnose evolving networks without committing to a single fixed mechanism, extending Bayesian predictive synthesis to the network setting with explicit uncertainty quantification and mechanism tracking.

major comments (1)
  1. [Abstract] Abstract: the central claims concerning a sharp identification threshold, optimal per-switch tracking cost, and single-graph weight recovery are stated without any derivation, proof sketch, or statement of the sparse-safe parametrization; these load-bearing theoretical results cannot be assessed for internal consistency or hidden circularity from the supplied text.
minor comments (1)
  1. [Abstract] The abstract sentence is truncated at 'recovers the leading mechanism when'.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their report. The sole major comment concerns the level of detail in the abstract regarding the theoretical claims. We respond point-by-point below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claims concerning a sharp identification threshold, optimal per-switch tracking cost, and single-graph weight recovery are stated without any derivation, proof sketch, or statement of the sparse-safe parametrization; these load-bearing theoretical results cannot be assessed for internal consistency or hidden circularity from the supplied text.

    Authors: We agree that the abstract presents the main results concisely without derivations, a proof sketch, or an explicit definition of the sparse-safe parametrization. This is standard for abstracts given length constraints, but we recognize it limits immediate assessability. The full development appears in the manuscript: the sparse-safe parametrization is introduced in Section 2.3, the identification theory with the sharp threshold is stated and proved in Theorem 3.1 and its corollaries, the optimal per-switch tracking cost is derived in Proposition 4.1, and single-snapshot weight recovery is shown in Section 3.4. To address the concern, we will revise the abstract to include a brief inline statement of the sparse-safe parametrization and a parenthetical reference to the relevant sections for the proofs. This change will be made in the next version. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected from available text

full rationale

The abstract describes a synthesis layer combining mechanism agents with time-varying weights, an identification theory under sparse-safe parametrization, and reductions to calibrated link prediction, but provides no equations, fitted-parameter renamings, or self-citation chains that reduce any claimed prediction or identification result to its inputs by construction. No load-bearing steps can be quoted or exhibited as circular; the derivation chain is therefore treated as self-contained on the given material.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review based solely on incomplete abstract; full details on parameters, axioms, and entities unavailable.

axioms (1)
  • domain assumption Structural mechanisms can be treated as independent forecasting agents whose weights are identifiable from a single graph snapshot.
    Invoked in the description of the synthesis layer and identification theory.

pith-pipeline@v0.9.1-grok · 5720 in / 1082 out tokens · 24701 ms · 2026-06-26T14:44:19.580089+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

89 extracted references · 2 linked inside Pith

  1. [1]

    Journal of econometrics , volume=

    Dynamic Bayesian predictive synthesis in time series forecasting , author=. Journal of econometrics , volume=. 2019 , publisher=

    2019

  2. [2]

    1997 , publisher=

    Bayesian forecasting and dynamic models , author=. 1997 , publisher=

    1997

  3. [3]

    The Annals of Mathematical Statistics , volume=

    Limiting behavior of posterior distributions when the model is incorrect , author=. The Annals of Mathematical Statistics , volume=. 1966 , publisher=

    1966

  4. [4]

    Journal of the American statistical Association , volume=

    Strictly proper scoring rules, prediction, and estimation , author=. Journal of the American statistical Association , volume=. 2007 , publisher=

    2007

  5. [5]

    Sankhya A , volume=

    A limit theorem for scaled eigenvectors of random dot product graphs , author=. Sankhya A , volume=. 2016 , publisher=

    2016

  6. [6]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    A statistical interpretation of spectral embedding: the generalised random dot product graph , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2022 , publisher=

    2022

  7. [7]

    The Annals of Statistics , pages=

    Rate-optimal graphon estimation , author=. The Annals of Statistics , pages=. 2015 , publisher=

    2015

  8. [8]

    Journal of the american statistical association , volume=

    Latent space models for dynamic networks , author=. Journal of the american statistical association , volume=. 2015 , publisher=

    2015

  9. [9]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Statistical clustering of temporal networks through a dynamic stochastic block model , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2017 , publisher=

    2017

  10. [10]

    Spectral clustering in the dynamic stochastic block model , author=

  11. [11]

    IEEE Journal of Selected Topics in Signal Processing , volume=

    Dynamic stochastic blockmodels for time-evolving social networks , author=. IEEE Journal of Selected Topics in Signal Processing , volume=. 2014 , publisher=

    2014

  12. [12]

    Social networks , volume=

    Friends and neighbors on the web , author=. Social networks , volume=. 2003 , publisher=

    2003

  13. [13]

    Biometrika , volume=

    Nonparametric Bayes dynamic modelling of relational data , author=. Biometrika , volume=. 2014 , publisher=

    2014

  14. [14]

    Random Structures & Algorithms , volume=

    Concentration and regularization of random graphs , author=. Random Structures & Algorithms , volume=. 2017 , publisher=

    2017

  15. [15]

    Introduction to Nonparametric Estimation , pages=

    Nonparametric estimators , author=. Introduction to Nonparametric Estimation , pages=. 2008 , publisher=

    2008

  16. [16]

    Journal of the Royal Statistical Society: Series A (General) , volume=

    Present position and potential developments: Some personal views statistical theory the prequential approach , author=. Journal of the Royal Statistical Society: Series A (General) , volume=. 1984 , publisher=

    1984

  17. [17]

    Using stacking to average Bayesian predictive distributions (with discussion) , author=

  18. [18]

    Journal of the american Statistical association , volume=

    Latent space approaches to social network analysis , author=. Journal of the american Statistical association , volume=. 2002 , publisher=

    2002

  19. [19]

    Social networks , volume=

    Stochastic blockmodels: First steps , author=. Social networks , volume=. 1983 , publisher=

    1983

  20. [20]

    Physical Review E—Statistical, Nonlinear, and Soft Matter Physics , volume=

    Stochastic blockmodels and community structure in networks , author=. Physical Review E—Statistical, Nonlinear, and Soft Matter Physics , volume=. 2011 , publisher=

    2011

  21. [21]

    Annals of combinatorics , volume=

    Connected components in random graphs with given expected degree sequences , author=. Annals of combinatorics , volume=. 2002 , publisher=

    2002

  22. [22]

    The Annals of Statistics , pages=

    Consistency of spectral clustering in stochastic block models , author=. The Annals of Statistics , pages=. 2015 , publisher=

    2015

  23. [23]

    Journal of Complex Networks , volume=

    Learning latent block structure in weighted networks , author=. Journal of Complex Networks , volume=. 2015 , publisher=

    2015

  24. [24]

    Proceedings of the twelfth international conference on Information and knowledge management , pages=

    The link prediction problem for social networks , author=. Proceedings of the twelfth international conference on Information and knowledge management , pages=

  25. [25]

    Clyde, David Draper and EI George, and a rejoinder by the authors , author=

    Bayesian model averaging: a tutorial (with comments by M. Clyde, David Draper and EI George, and a rejoinder by the authors , author=. Statistical science , volume=. 1999 , publisher=

    1999

  26. [26]

    Neural networks , volume=

    Stacked generalization , author=. Neural networks , volume=. 1992 , publisher=

    1992

  27. [27]

    1991 , publisher=

    Modelling with mixtures , author=. 1991 , publisher=

    1991

  28. [28]

    Limit theorems for eigenvectors of the normalized Laplacian for random graphs , author=

  29. [29]

    Statistics surveys , volume=

    A review of dynamic network models with latent variables , author=. Statistics surveys , volume=

  30. [30]

    arXiv preprint arXiv:2512.18584 , year=

    State-Space Modeling of Time-Varying Spillovers on Networks , author=. arXiv preprint arXiv:2512.18584 , year=

    arXiv

  31. [31]

    arXiv preprint arXiv:2512.18587 , year=

    Graphon-Level Bayesian Predictive Synthesis for Random Network , author=. arXiv preprint arXiv:2512.18587 , year=

    arXiv

  32. [32]

    arXiv preprint arXiv:2606.15836 , year=

    Minimax Synthesis of Network Mechanisms , author=. arXiv preprint arXiv:2606.15836 , year=

    arXiv

  33. [33]

    Econometrica: Journal of the econometric society , pages=

    Maximum likelihood estimation of misspecified models , author=. Econometrica: Journal of the econometric society , pages=. 1982 , publisher=

    1982

  34. [34]

    Political analysis , volume=

    Cluster--robust variance estimation for dyadic data , author=. Political analysis , volume=. 2015 , publisher=

    2015

  35. [35]

    The Annals of Probability , pages=

    Normal convergence by higher semiinvariants with applications to sums of dependent random variables and random graphs , author=. The Annals of Probability , pages=. 1988 , publisher=

    1988

  36. [36]

    The Annals of Statistics , pages=

    Convergence of estimates under dimensionality restrictions , author=. The Annals of Statistics , pages=. 1973 , publisher=

    1973

  37. [37]

    Biometrika , volume=

    Logistic disease incidence models and case-control studies , author=. Biometrika , volume=. 1979 , publisher=

    1979

  38. [38]

    Statistical science: a review journal of the Institute of Mathematical Statistics , volume=

    On the question of effective sample size in network modeling: An asymptotic inquiry , author=. Statistical science: a review journal of the Institute of Mathematical Statistics , volume=

  39. [39]

    Journal of the Royal Statistical Society Series A: Statistics in Society , volume=

    On the reconciliation of probability assessments , author=. Journal of the Royal Statistical Society Series A: Statistics in Society , volume=. 1979 , publisher=

    1979

  40. [40]

    Journal of the Royal Statistical Society: Series A (General) , volume=

    Bayesian aggregation , author=. Journal of the Royal Statistical Society: Series A (General) , volume=. 1984 , publisher=

    1984

  41. [41]

    Technometrics , volume=

    Online prediction under model uncertainty via dynamic model averaging: Application to a cold rolling mill , author=. Technometrics , volume=. 2010 , publisher=

    2010

  42. [42]

    Nature Communications , volume=

    Sequential stacking link prediction algorithms for temporal networks , author=. Nature Communications , volume=. 2024 , publisher=

    2024

  43. [43]

    Misspecification in infinite-dimensional Bayesian statistics , author=

  44. [44]

    Oracle inequalities for network models and sparse graphon estimation , author=

  45. [45]

    Handbook of econometrics , volume=

    Network data , author=. Handbook of econometrics , volume=. 2020 , publisher=

    2020

  46. [46]

    Proceedings of the National Academy of Sciences , volume=

    Stacking models for nearly optimal link prediction in complex networks , author=. Proceedings of the National Academy of Sciences , volume=. 2020 , publisher=

    2020

  47. [47]

    Journal of Machine Learning Research , volume=

    Community detection and stochastic block models: recent developments , author=. Journal of Machine Learning Research , volume=

  48. [48]

    Physical Review E—Statistical, Nonlinear, and Soft Matter Physics , volume=

    Asymptotic analysis of the stochastic block model for modular networks and its algorithmic applications , author=. Physical Review E—Statistical, Nonlinear, and Soft Matter Physics , volume=. 2011 , publisher=

    2011

  49. [49]

    Probability Theory and Related Fields , volume=

    Reconstruction and estimation in the planted partition model , author=. Probability Theory and Related Fields , volume=. 2015 , publisher=

    2015

  50. [50]

    Random Structures & Algorithms , volume=

    Testing for high-dimensional geometry in random graphs , author=. Random Structures & Algorithms , volume=. 2016 , publisher=

    2016

  51. [51]

    Journal of the American Statistical Association , volume=

    Multivariate Bayesian predictive synthesis in macroeconomic forecasting , author=. Journal of the American Statistical Association , volume=. 2020 , publisher=

    2020

  52. [52]

    Econometric Theory , pages=

    DOUBLE/DEBIASED MACHINE LEARNING FOR DYADIC DATA , author=. Econometric Theory , pages=. 2026 , publisher=

    2026

  53. [53]

    arXiv preprint arXiv:1710.00862 , year=

    Testing for global network structure using small subgraph statistics , author=. arXiv preprint arXiv:1710.00862 , year=

    Pith/arXiv arXiv

  54. [54]

    International conference on machine learning , pages=

    Network global testing by counting graphlets , author=. International conference on machine learning , pages=. 2018 , organization=

    2018

  55. [55]

    2016 IEEE 16th international conference on data mining (ICDM) , pages=

    Edge weight prediction in weighted signed networks , author=. 2016 IEEE 16th international conference on data mining (ICDM) , pages=. 2016 , organization=

    2016

  56. [56]

    ACM Transactions on Intelligent Systems and Technology (TIST) , volume=

    Snap: A general-purpose network analysis and graph-mining library , author=. ACM Transactions on Intelligent Systems and Technology (TIST) , volume=. 2016 , publisher=

    2016

  57. [57]

    ACM transactions on Knowledge Discovery from Data (TKDD) , volume=

    Graph evolution: Densification and shrinking diameters , author=. ACM transactions on Knowledge Discovery from Data (TKDD) , volume=. 2007 , publisher=

    2007

  58. [58]

    nature , volume=

    Collective dynamics of ‘small-world’networks , author=. nature , volume=. 1998 , publisher=

    1998

  59. [59]

    Proceedings of the 3rd international workshop on Link discovery , pages=

    The political blogosphere and the 2004 US election: divided they blog , author=. Proceedings of the 3rd international workshop on Link discovery , pages=

    2004

  60. [60]

    arXiv preprint arXiv:2505.22289 , year=

    Network Model Averaging Prediction for Latent Space Models by K-Fold Edge Cross-Validation , author=. arXiv preprint arXiv:2505.22289 , year=

    arXiv

  61. [61]

    Advances in neural information processing systems , volume=

    Mixed membership stochastic blockmodels , author=. Advances in neural information processing systems , volume=

  62. [62]

    Biometrika , volume=

    Network cross-validation by edge sampling , author=. Biometrika , volume=. 2020 , publisher=

    2020

  63. [63]

    The Annals of Statistics , pages=

    Modeling expert judgments for Bayesian updating , author=. The Annals of Statistics , pages=. 1985 , publisher=

    1985

  64. [64]

    Statistical Science , volume=

    Combining probability distributions: A critique and an annotated bibliography , author=. Statistical Science , volume=. 1986 , publisher=

    1986

  65. [65]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Modelling probabilistic agent opinion , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 1992 , publisher=

    1992

  66. [66]

    Journal of the Royal Statistical Society Series C: Applied Statistics , volume=

    Mixed-frequency Bayesian predictive synthesis for economic nowcasting , author=. Journal of the Royal Statistical Society Series C: Applied Statistics , volume=. 2021 , publisher=

    2021

  67. [67]

    Journal of Econometrics , volume=

    Optimal prediction pools , author=. Journal of Econometrics , volume=. 2011 , publisher=

    2011

  68. [68]

    Physica A: statistical mechanics and its applications , volume=

    Link prediction in complex networks: A survey , author=. Physica A: statistical mechanics and its applications , volume=. 2011 , publisher=

    2011

  69. [69]

    Foundations and Trends

    A survey of statistical network models , author=. Foundations and Trends. 2010 , publisher=

    2010

  70. [70]

    arXiv preprint arXiv:2305.06465 , year=

    Occam Factor for Random Graphs: Erd " \ o \ sR ' \ e \ nyi, Independent Edge, and Rank-1 Stochastic Blockmodel , author=. arXiv preprint arXiv:2305.06465 , year=

    arXiv

  71. [71]

    Journal of the Royal Statistical Society Series C: Applied Statistics , volume=

    Analysis of the formation of the structure of social networks by using latent space models for ranked dynamic networks , author=. Journal of the Royal Statistical Society Series C: Applied Statistics , volume=. 2015 , publisher=

    2015

  72. [72]

    Discrete temporal models of social networks , author=

  73. [73]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    A separable model for dynamic networks , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2014 , publisher=

    2014

  74. [74]

    Journal of Multimedia Information System , volume=

    A Growing Stochastic Block Model with Preferential Attachment , author=. Journal of Multimedia Information System , volume=. 2025 , publisher=

    2025

  75. [75]

    Journal of the Royal Statistical Society Series A: Statistics in Society , volume=

    Zero-inflated stochastic block modelling of efficiency-security trade-offs in weighted criminal networks , author=. Journal of the Royal Statistical Society Series A: Statistics in Society , volume=. 2026 , publisher=

    2026

  76. [76]

    Journal of Complex Networks , volume=

    Scaling laws for properties of random graphs that grow via successive combination , author=. Journal of Complex Networks , volume=. 2022 , publisher=

    2022

  77. [77]

    Symmetry , volume=

    A novel algorithm for merging Bayesian networks , author=. Symmetry , volume=. 2023 , publisher=

    2023

  78. [78]

    2017 , school=

    Bayesian predictive synthesis: Forecast calibration and combination , author=. 2017 , school=

    2017

  79. [79]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Bayesian predictive decision synthesis , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2024 , publisher=

    2024

  80. [80]

    arXiv preprint arXiv:2311.12671 , year=

    Predictive density combination using a tree-based synthesis function , author=. arXiv preprint arXiv:2311.12671 , year=

    arXiv

Showing first 80 references.