pith. sign in

arxiv: 2507.09983 · v2 · pith:PMC6DM22new · submitted 2025-07-14 · 📊 stat.AP · stat.ME

Gradient boosted multi-population mortality modelling with high-frequency data

Ziting Miao , Han Li , Yuyu Chen This is my paper

Pith reviewed 2026-05-19 05:25 UTC · model grok-4.3

classification 📊 stat.AP stat.ME
keywords mortality modellinggradient boostingmulti-populationhigh-frequency dataseasonal patternsLi-Lee modelforecast accuracyclustering
0
0 comments X

The pith

Embedding the Li and Lee model as a weak learner in gradient boosting improves fit and forecast accuracy for weekly multi-population mortality data.

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

This paper shows that traditional stochastic mortality models gain from being placed inside a gradient boosting framework when the data arrive at weekly frequency. The authors substitute the Li and Lee model for ordinary decision trees so that each boosting step corrects residuals while preserving the multi-population structure of trends and age effects. If the claim holds, demographers and insurers could obtain short-term forecasts that better reflect seasonal swings without first discarding or heavily smoothing the high-frequency observations. The study tests the idea on weekly records from thirty countries and reports gains in both in-sample fit and out-of-sample accuracy over standard benchmarks.

Core claim

The central claim is that replacing conventional decision trees with the Li and Lee model inside a gradient boosting loop, applied in a multi-population setting, produces mortality models that capture both long-term trends and short-term seasonal fluctuations more accurately than existing approaches, yielding superior forecast performance on weekly data from thirty countries while remaining stable across different ways of grouping the populations.

What carries the argument

Gradient boosting iteration that treats the Li and Lee multi-population model as the weak learner, allowing successive residual corrections while retaining the stochastic age-period-cohort structure.

If this is right

  • Weekly mortality series can be modelled directly without first aggregating to annual frequency or removing seasonal components.
  • Forecast accuracy improves relative to benchmark stochastic models and to gradient boosting that uses decision trees.
  • Model performance stays high across multiple choices of how to cluster countries into coherent sub-populations.
  • The need for extensive preliminary data cleaning or population selection is reduced.

Where Pith is reading between the lines

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

  • The same boosting construction could be tried on other high-frequency vital statistics such as weekly birth counts or hospital admissions.
  • The clustering step based on improvement rates and seasonal strength offers a template for grouping coherent units in related demographic or epidemiological time series.
  • Extending the method to include time-varying covariates such as temperature or economic indicators would be a direct next test.

Load-bearing premise

The Li and Lee model, built for annual series with long-run cohort effects, still works well as a weak learner when the dominant signal in the data is short-term weekly seasonality.

What would settle it

A direct comparison in which the proposed boosted model produces higher out-of-sample mean squared forecast errors than either a standard Li and Lee model or ordinary tree-based gradient boosting on weekly mortality series from countries or years held out of the 2015-2019 sample.

Figures

Figures reproduced from arXiv: 2507.09983 by Han Li, Yuyu Chen, Ziting Miao.

Figure 1
Figure 1. Figure 1: 2015–2019 mortality rates for the four representative countries by age groups [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The residual plot of Canada under the LL model [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The residual plot of Canada under the HBY model [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Components of the common trends under the GBLL model: first iteration (left) and [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Improvement in MAPE from LL to GBLL (left) and from HBY to GBLL (right) [PITH_FULL_IMAGE:figures/full_fig_p015_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Mean value of the time trends in each cluster [PITH_FULL_IMAGE:figures/full_fig_p017_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Trend slopes by clusters All countries have negative trend slopes, indicating the mortality improvement across five years. The rapid mortality improvement observed in Cluster 1 may be attributed to factors such as healthcare system reforms and advances in medical technology and treatment (see, e.g., OECD, 2021; Pekarcikova, 2024). Given that countries like Croatia, Lithuania, and Slovakia have histor￾icall… view at source ↗
Figure 8
Figure 8. Figure 8: Clusters based on min-max scaled trend slopes and seasonal strength [PITH_FULL_IMAGE:figures/full_fig_p019_8.png] view at source ↗
read the original abstract

High-frequency mortality data have attracted growing attention, but their use has largely been confined to specific applications rather than general modelling and forecasting. Such data pose new challenges to traditional mortality models due to pronounced seasonal patterns and short-term fluctuations. To address these challenges and produce more accurate forecasts with the high-frequency mortality data, this paper introduces a novel integration of gradient boosting techniques into traditional stochastic mortality models under a multi-population setting. Our key innovation lies in using the Li and Lee model as the weak learner within the gradient boosting framework, replacing conventional decision trees. Empirical studies are conducted using weekly mortality data from 30 countries (Human Mortality Database, 2015-2019). Empirical evidence highlights that the proposed methodology not only enhances model fit by accurately capturing underlying mortality trends and seasonal patterns, but also achieves superior forecast accuracy, compared to the benchmark models. We also investigate a key challenge in multi-population mortality modelling: how to select appropriate sub-populations with sufficiently similar mortality experiences. A comprehensive clustering exercise is conducted based on mortality improvement rates and seasonal strength. The empirical results demonstrate that our proposed model maintains strong forecast accuracy across different clustering configurations, thereby reducing the need for extensive data preprocessing.

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

2 major / 1 minor

Summary. The paper proposes integrating gradient boosting into multi-population stochastic mortality models by using the Li and Lee model as the weak learner instead of decision trees. Applied to weekly mortality data from 30 countries (2015-2019, Human Mortality Database), it claims improved fit by capturing trends and seasonal patterns, superior forecast accuracy versus benchmarks, and robustness of results across different sub-population clustering configurations based on mortality improvement rates and seasonal strength.

Significance. If the empirical advantages are confirmed with quantitative detail, the work could advance high-frequency mortality modeling by adapting established age-period-cohort structures to short-term data, with potential value for actuarial forecasting and public health applications that require timely seasonal adjustments.

major comments (2)
  1. Abstract (key innovation paragraph): The central claim rests on the Li and Lee model serving effectively as a weak learner for weekly data whose dominant features are short-term seasonality and noise rather than the long-run trends and cohorts for which it was derived. The abstract states that seasonal patterns are captured but does not indicate whether an explicit seasonal term was added to the Li-Lee specification or whether performance is robust to alternative base learners; without this, gains may be driven by the boosting machinery itself, weakening the specific methodological contribution.
  2. Empirical studies section: The claim of superior forecast accuracy is presented without reference to specific quantitative metrics (e.g., MAE, RMSE, or log-likelihood differences), cross-validation procedures, or ablation results against the benchmarks. This makes the magnitude and reliability of the reported improvements difficult to evaluate and is load-bearing for the paper's primary empirical conclusion.
minor comments (1)
  1. Clustering exercise: Additional detail on the distance metric, linkage method, and criteria for determining 'sufficiently similar mortality experiences' would aid reproducibility of the sub-population selection results.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their detailed and constructive comments on our manuscript. We have addressed each major comment below and will make revisions to improve the clarity and rigor of the paper.

read point-by-point responses

  1. Referee: Abstract (key innovation paragraph): The central claim rests on the Li and Lee model serving effectively as a weak learner for weekly data whose dominant features are short-term seasonality and noise rather than the long-run trends and cohorts for which it was derived. The abstract states that seasonal patterns are captured but does not indicate whether an explicit seasonal term was added to the Li-Lee specification or whether performance is robust to alternative base learners; without this, gains may be driven by the boosting machinery itself, weakening the specific methodological contribution.

    Authors: We appreciate this observation. Our methodology employs the standard Li-Lee model as the weak learner without incorporating an additional explicit seasonal term. The gradient boosting procedure enables the capture of seasonal patterns by successively fitting the Li-Lee model to the residuals from previous iterations, which is particularly effective for the short-term fluctuations in weekly mortality data. This adaptation is central to our contribution. We did not perform comparisons with alternative base learners such as decision trees in the present study. In the revision, we will update the abstract to explicitly state that no seasonal term is added to the Li-Lee specification and provide a brief justification for selecting the Li-Lee model in the multi-population context. We will also add a sentence noting that exploring alternative weak learners remains an avenue for future research. revision: partial

  2. Referee: Empirical studies section: The claim of superior forecast accuracy is presented without reference to specific quantitative metrics (e.g., MAE, RMSE, or log-likelihood differences), cross-validation procedures, or ablation results against the benchmarks. This makes the magnitude and reliability of the reported improvements difficult to evaluate and is load-bearing for the paper's primary empirical conclusion.

    Authors: We agree that quantitative details are crucial for substantiating the claims. The full manuscript includes comparisons using metrics such as mean absolute error (MAE) and root mean squared error (RMSE) for forecast accuracy, as well as a description of the time-series cross-validation approach used. To address the referee's concern, we will revise the empirical studies section to more prominently feature these specific metrics, include detailed numerical results in tables, elaborate on the cross-validation procedure, and incorporate ablation analyses to better isolate the effects of the gradient boosting integration versus the base model. These changes will enhance the transparency and evaluability of our empirical findings. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical performance claims rest on external benchmarks rather than self-referential derivation

full rationale

The paper proposes a gradient-boosting framework that substitutes the Li-Lee model for decision trees as the weak learner and then reports improved in-sample fit and out-of-sample forecast accuracy on weekly mortality data from 30 countries. These performance claims are evaluated against separately implemented benchmark models and are therefore falsifiable on held-out data; they do not reduce by construction to any fitted parameter, self-citation, or redefinition of the target quantity. The clustering step for subpopulation selection is likewise an independent empirical exercise. No load-bearing equation or uniqueness theorem is shown to collapse into its own inputs, satisfying the default expectation that an empirical methodological paper contains no significant circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on the domain assumption that the Li-Lee model structure can serve as an effective base learner for short-term seasonal corrections and on the empirical choice of clustering variables; no new free parameters or invented entities are introduced in the abstract.

axioms (1)
  • domain assumption Li-Lee model remains a suitable weak learner for weekly mortality series dominated by seasonality
    Invoked in the description of the key innovation; if false, the boosting steps would not systematically reduce seasonal residuals.

pith-pipeline@v0.9.0 · 5736 in / 1234 out tokens · 31675 ms · 2026-05-19T05:25:09.110915+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

61 extracted references · 61 canonical work pages

  1. [1]

    G., and Tragaki, A

    Andreopoulos, P., Bersimis, F. G., and Tragaki, A. (2022). A different approach to current de- velopments in the twenty-first century–grouping european countries in terms of mortality. In Quantitative Methods in Demography: Methods and Related Applications in the Covid-19 Era , pages 373–385. Springer

    work page 2022

  2. [2]

    Bjerre, D. S. (2022). Tree-based machine learning methods for modeling and forecasting mortality. ASTIN Bulletin: The Journal of the IAA , 52(3):765–787

    work page 2022

  3. [3]

    Bonnet, C., Cambois, E., Fontaine, R., Dutreuilh, C., and van Hoorn Alkema, B. (2021). Popu- lation ageing in high-longevity countries: demographic dynamics and socio-economic challenges. Population, 76(2):217–310

    work page 2021

  4. [4]

    Boonen, T. J. and Chen, Y. (2025). Low-rank tensor autoregressive models for mortality modeling. Available at SSRN 5233127

    work page 2025

  5. [5]

    Brouhns, N., Denuit, M., and Vermunt, J. K. (2002). A poisson log-bilinear regression approach to the construction of projected lifetables. Insurance: Mathematics and Economics , 31(3):373–393

    work page 2002

  6. [6]

    J., Blake, D., and Dowd, K

    Cairns, A. J., Blake, D., and Dowd, K. (2006). A two-factor model for stochastic mortality with parameter uncertainty: theory and calibration. Journal of Risk and Insurance , 73(4):687–718

    work page 2006

  7. [7]

    and Guestrin, C

    Chen, T. and Guestrin, C. (2016). Xgboost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , pages 785–794. 24

    work page 2016

  8. [8]

    B., Cleveland, W

    Cleveland, R. B., Cleveland, W. S., McRae, J. E., Terpenning, I., et al. (1990). STL: A seasonal- trend decomposition. Journal of Official Statistics , 6(1):3–73

    work page 1990

  9. [9]

    L., Haberman, S., and Millossovich, P

    Danesi, I. L., Haberman, S., and Millossovich, P. (2015). Forecasting mortality in subpopulations using Lee–Carter type models: A comparison. Insurance: Mathematics and Economics , 62, 151–161

    work page 2015

  10. [10]

    Debon, A., Haberman, S., and Piscopo, G. (2024). Multipopulation mortality analysis: bringing out the unobservable with latent clustering. Quality & Quantity , 58(6):5107–5123. Demirta¸ s, M. (2022). The anomalously cold January 2017 in the south-eastern europe in a warming climate. International Journal of Climatology , 42(11):6018–6026

    work page 2024

  11. [11]

    V., and W¨ uthrich, M

    Deprez, P., Shevchenko, P. V., and W¨ uthrich, M. V. (2017). Machine learning techniques for mortality modeling. European Actuarial Journal, 7, 337–352

    work page 2017

  12. [12]

    Dimai, M. (2025). Multi-population mortality modeling with economic, environmental and lifestyle variables. Quality & Quantity , 59, 153–205

    work page 2025

  13. [13]

    E., Karageorgopoulos, D

    Falagas, M. E., Karageorgopoulos, D. E., Moraitis, L. I., Vouloumanou, E. K., Roussos, N., Pep- pas, G., and Rafailidis, P. I. (2009). Seasonality of mortality: the September phenomenon in Mediterranean countries. Canadian Medical Association Journal , 181(8):484–486

    work page 2009

  14. [14]

    Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. Annals of Statistics, 29(5):1189–1232

    work page 2001

  15. [15]

    K., Goswami, S., and Panday, M

    Ghosal, A., Nandy, A., Das, A. K., Goswami, S., and Panday, M. (2020). A short review on different clustering techniques and their applications. Emerging Technology in Modelling and Graphics: Proceedings of IEM Graph 2018, pages 69–83

    work page 2020

  16. [16]

    and King, G

    Girosi, F. and King, G. (2007). Understanding the Lee–Carter mortality forecasting method. Tech- nical report, RAND Corporation

    work page 2007

  17. [17]

    Guelman, L. (2012). Gradient boosting trees for auto insurance loss cost modeling and prediction. Expert Systems with Applications , 39(3):3659–3667

    work page 2012

  18. [18]

    and Haberman, S

    Hatzopoulos, P. and Haberman, S. (2013). Common mortality modeling and coherent forecasts. An empirical analysis of worldwide mortality data. Insurance: Mathematics and Economics , 52(2):320–337

    work page 2013

  19. [19]

    Huang, C., Chu, C., Wang, X., and Barnett, A. G. (2015). Unusually cold and dry winters increase mortality in Australia. Environmental Research, 136, 1–7

    work page 2015

  20. [20]

    J., Booth, H., and Yasmeen, F

    Hyndman, R. J., Booth, H., and Yasmeen, F. (2013). Coherent mortality forecasting: the product- ratio method with functional time series models. Demography, 50(1):261–283

    work page 2013

  21. [21]

    Hyndman, R. J. and Ullah, M. S. (2007). Robust forecasting of mortality and fertility rates: A functional data approach. Computational Statistics & Data Analysis , 51(10):4942–4956

    work page 2007

  22. [22]

    M., Acosta, R

    Islam, N., Shkolnikov, V. M., Acosta, R. J., Klimkin, I., Kawachi, I., Irizarry, R. A., Alicandro, G., Khunti, K., Yates, T., Jdanov, D. A., et al. (2021). Excess deaths associated with covid-19 pandemic in 2020: age and sex disaggregated time series analysis in 29 high income countries. BMJ, 373:n1137

    work page 2021

  23. [23]

    Jacobsen, R., Keiding, N., and Lynge, E. (2002). Long term mortality trends behind low life expectancy of danish women. Journal of Epidemiology & Community Health , 56(3):205–208

    work page 2002

  24. [24]

    A., Galarza, A

    Jdanov, D. A., Galarza, A. A., Shkolnikov, V. M., Jasilionis, D., N´ emeth, L., Leon, D. A., Boe, C., and Barbieri, M. (2021). The short-term mortality fluctuation data series, monitoring mortality shocks across time and space. Scientific Data, 8(1):235

    work page 2021

  25. [25]

    and Kobak, D

    Karlinsky, A. and Kobak, D. (2021). The world mortality dataset: Tracking excess mortality across countries during the covid-19 pandemic. eLife, 10:e69336

    work page 2021

  26. [26]

    Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q., and Liu, T. (2017). Lightgbm: A highly efficient gradient boosting decision tree. Proceedings of the 31st International Conference 25 on Neural Information Processing Systems , pages 3149–3157

    work page 2017

  27. [27]

    E., Collier, T., Dunstone, N., Gordon, M., Hardiman, S., Her- manson, L., Ineson, S., Kay, G., et al

    Knight, J., Scaife, A., Bett, P. E., Collier, T., Dunstone, N., Gordon, M., Hardiman, S., Her- manson, L., Ineson, S., Kay, G., et al. (2021). Predictability of european winters 2017/2018 and 2018/2019: Contrasting influences from the tropics and stratosphere. Atmospheric Science Letters, 22(1):e1009

    work page 2021

  28. [28]

    Zhou, B., Battaglini, M., Corsetti, G., et al. (2020). Magnitude, demographics and dynamics of the effect of the first wave of the covid-19 pandemic on all-cause mortality in 21 industrialized countries. Nature Medicine, 26(12):1919–1928

    work page 2020

  29. [29]

    Kostopoulou, E. (2023). Analysis of the January 2017 cold spell in greece and its implications on human health. Environmental Sciences Proceedings, 26(1):195

    work page 2023

  30. [30]

    Lam, K. K. and Wang, B. (2023). Multipopulation mortality modelling and forecasting: the weighted multivariate functional principal component approaches. Journal of Applied Statistics , 50(15):3177–3198

    work page 2023

  31. [31]

    Lawrence, R., Bunn, A., Powell, S., and Zambon, M. (2004). Classification of remotely sensed imagery using stochastic gradient boosting as a refinement of classification tree analysis. Remote Sensing of Environment , 90(3):331–336

    work page 2004

  32. [32]

    Lee, R. D. and Carter, L. R. (1992). Modeling and forecasting U.S. mortality. Journal of the American Statistical Association, 87(419):659–671. L´ eger, A. E. and Mazzuco, S. (2021). What can we learn from the functional clustering of mortality data? An application to the human mortality database. European Journal of Population , 37, 769–798. L´ eger, A. ...

    work page 1992

  33. [33]

    and Pizzorusso, V

    Levantesi, S. and Pizzorusso, V. (2019). Application of machine learning to mortality modeling and forecasting. Risks, 7(1):26

    work page 2019

  34. [34]

    and Chen, H

    Li, H. and Chen, H. (2024). Hierarchical mortality forecasting with EVT tails: An application to solvency capital requirement. International Journal of Forecasting, 40(2):549–563

    work page 2024

  35. [35]

    Li, H., Li, H., Lu, Y., and Panagiotelis, A. (2019). A forecast reconciliation approach to cause-of- death mortality modeling. Insurance: Mathematics and Economics , 86, 122–133

    work page 2019

  36. [36]

    and Tang, Q

    Li, H. and Tang, Q. (2022). Joint extremes in temperature and mortality: A bivariate POT approach. North American Actuarial Journal , 26(1):43–63

    work page 2022

  37. [37]

    Li, L., Li, H., and Panagiotelis, A. (2025). Boosting domain-specific models with shrinkage: an application in mortality forecasting. International Journal of Forecasting, 41(1):191–207

    work page 2025

  38. [38]

    and Lee, R

    Li, N. and Lee, R. (2005). Coherent mortality forecasts for a group of populations: An extension of the Lee–Carter method. Demography, 42, 575–594

    work page 2005

  39. [39]

    Ljung, G. M. and Box, G. E. (1978). On a measure of lack of fit in time series models. Biometrika, 65(2):297–303

    work page 1978

  40. [40]

    Lloyd, S. (1982). Least squares quantization in PCM. IEEE Transactions on Information Theory , 28(2):129–137

    work page 1982

  41. [41]

    and Rogers, A

    McNown, R. and Rogers, A. (1992). Forecasting cause-specific mortality using time series methods. International Journal of Forecasting, 8(3):413–432

    work page 1992

  42. [42]

    and Russo, A

    Murtas, R. and Russo, A. G. (2019). Effects of pollution, low temperature and influenza syndrome on the excess mortality risk in winter 2016–2017. BMC Public Health , 19, 1–9

    work page 2019

  43. [43]

    and Knoll, A

    Natekin, A. and Knoll, A. (2013). Gradient boosting machines, a tutorial. Frontiers in Neuro- robotics, 7:21

    work page 2013

  44. [44]

    R., Klimkin, I., Jdanov, D

    Nepomuceno, M. R., Klimkin, I., Jdanov, D. A., Alustiza-Galarza, A., and Shkolnikov, V. M. 26 (2022). Sensitivity analysis of excess mortality due to the covid-19 pandemic. Population and Development Review, 48(2):279–302

    work page 2022

  45. [45]

    Neves, C., Fernandes, C., and Hoeltgebaum, H. (2017). Five different distributions for the Lee– Carter model of mortality forecasting: A comparison using GAS models. Insurance: Mathematics and Economics, 75, 48–57. OECD (2021). State of Health in the EU Croatia: Country Health Profile 2021 . OECD Publishing

    work page 2017

  46. [46]

    Osmond, C. (1985). Using age, period and cohort models to estimate future mortality rates. International Journal of Epidemiology , 14(1):124–129

    work page 1985

  47. [47]

    Pekarcikova, J. (2024). Cancer mortality attributable to air pollution in slovakia. European Journal of Public Health , 34(Supplement 3):ckae144.1429

    work page 2024

  48. [48]

    Qiao, Y., Wang, C., and Zhu, W. (2024). Machine learning in long-term mortality forecasting. The Geneva Papers on Risk and Insurance – Issues and Practice , 49(2):340–362

    work page 2024

  49. [49]

    Renshaw, A. E. and Haberman, S. (2006). A cohort-based extension to the Lee–Carter model for mortality reduction factors. Insurance: Mathematics and Economics , 38(3):556–570

    work page 2006

  50. [50]

    and Maimon, O

    Rokach, L. and Maimon, O. (2005). Clustering methods. Data Mining and Knowledge Discovery Handbook, pages 321–352

    work page 2005

  51. [51]

    Rushin, G., Stancil, C., Sun, M., Adams, S., and Beling, P. (2017). Horse race analysis in credit card fraud—deep learning, logistic regression, and gradient boosted tree. In 2017 Systems and Information Engineering Design Symposium (SIEDS) , pages 117–121. IEEE

    work page 2017

  52. [52]

    Russolillo, M., Giordano, G., and Haberman, S. (2011). Extending the Lee–Carter model: a three- way decomposition. Scandinavian Actuarial Journal, 2011(2):96–117

    work page 2011

  53. [53]

    Serfling, R. E. (1963). Methods for current statistical analysis of excess pneumonia-influenza deaths. Public Health Reports, 78(6):494

    work page 1963

  54. [54]

    Shapovalov, V., Landsman, Z., and Makov, U. (2019). Bayesian log-bilinear mortality projection with a random walk with drift. Available at SSRN 3375920 . STMF (2021). Short-Term Mortality Fluctuation Data Series. Human Mortality Database. https: //www.mortality.org/Data/STMF

    work page 2019

  55. [55]

    and Bai, M

    Su, X. and Bai, M. (2020). Stochastic gradient boosting frequency-severity model of insurance claims. PLOS ONE, 15(8):e0238000

    work page 2020

  56. [56]

    Thorndike, R. L. (1953). Who belongs in the family? Psychometrika, 18(4):267–276

    work page 1953

  57. [57]

    Tian, Z., Xiao, J., Feng, H., and Wei, Y. (2020). Credit risk assessment based on gradient boosting decision tree. Procedia Computer Science, 174, 150–160

    work page 2020

  58. [58]

    Tsai, C. C.-L. and Cheng, E. S. (2021). Incorporating statistical clustering methods into mortality models to improve forecasting performances. Insurance: Mathematics and Economics, 99, 42–62

    work page 2021

  59. [59]

    Vanella, P., Basellini, U., and Lange, B. (2021). Assessing excess mortality in times of pandemics based on principal component analysis of weekly mortality data—the case of covid-19. Genus, 77, 1–36

    work page 2021

  60. [60]

    Yin, H., Aryani, A., Petrie, S., Nambissan, A., Astudillo, A., and Cao, S. (2024). A rapid review of clustering algorithms. arXiv preprint arXiv:2401.07389

    work page arXiv 2024

  61. [61]

    Zhou, J., Shi, X., Huang, R., Qiu, X., and Chen, C. (2016). Feasibility of stochastic gradient boost- ing approach for predicting rockburst damage in burst-prone mines. Transactions of Nonferrous Metals Society of China , 26(7):1938–1945. 27

    work page 2016