In generalized regression models, the effect of continuous covariates is commonly assumed to be linear. This assumption, however, may be too restrictive in applications and may lead to biased effect estimates and decreased predictive ability. While a multitude of alternatives for the flexible modelling of continuous covariates have been proposed, methods that provide guidance for choosing a suitable functional form are still limited. To address this issue, we propose a detection algorithm that evaluates several approaches for modelling continuous covariates and guides practitioners to choose the most appropriate alternative. The algorithm utilizes a unified framework for tree-structured modelling which makes the results easily interpretable. We assessed the performance of the algorithm by conducting a simulation study. To illustrate the proposed algorithm, we analysed data of patients suffering from chronic kidney disease.
AkaikeH (1974) A new look at the statistical model identification. IEEE Transactions on Automatic Control, 19, 716–23. doi: 10.1109/TAC.1974.1100705
2.
AndersenR (2009) Nonparametric methods for modeling nonlinearity in regression analysis. Annual Review of Sociology, 35, 67–85. doi: 10.1146/annurev.soc.34.040507.134631
3.
AndreouE and GhyselsE (2009) Structural breaks in financial time series. In: Handbook of Financial Time Series. Berlin Heidelberg: Springer pp. 839–70. doi: 10.1007/978-3-540-71297-8
4.
BergerM (2025) TSVC: Tree-structured modelling of varying coefficients. R package version 1.7.2. https://CRAN.R-project.org/package=TSVC
5.
BergerM, TutzG and SchmidM (2019) Tree-structured modelling of varying coefficients. Statistics & Computing, 29, 217–29. doi: 10.1007/s11222-018-9804-8
6.
BonatePL (2016) Effect of correlation on covariate selection in linear and nonlinear mixed effect models. Pharmaceutical Statistics, 16. doi: 10.1002/pst.1776
7.
BoulesteixAL, StroblC, AugustinT and DaumerM (2008) Evaluating microarray-based classifiers: An overview. Cancer Informatics, 6, 77–97. doi: 10.4137%2Fcin.s408
BreimanL, FriedmanJH, OlshenRA and StoneJC (1984) Classification and Regression Trees. Taylor and Francis, Monterey, CA: Wadsworth. doi: 10.1201/9781315139470
10.
CaoR, HuH and WanQ (2022) Association between hemoglobin and eGFR in patients with type 2 diabetes mellitus. Biomedical Journal of Scientific & Technical Research, 45.
11.
ChenY and YangY (2021) The one standard error rule for model selection: does it work?Stats, 4, 868–92. doi: 10.3390/stats4040051
12.
ChowGC (1960) Tests of equality between sets of coefficients in two linear regressions. Econometrica, 28, 591–605. doi: 10.2307/1910133
13.
De BoorC (1978) A Practical Guide to Splines. Springer, New York. doi: 10.1007/978-1-4612-6333-3
EilersPHC and MarxBD (1996) Flexible smoothing with B-splines and penalties. Statistical Science, 11, 89–121. doi: 10.1214/SS% 2F1038425655
16.
ElisseeffA, EvgeniouT and PontilM (2005) Stability of randomized learning algorithms. Journal of Machine Learning Research, 6, 55–79.
17.
EllenbachN, A-LBoulesteix and BischlB (2021) Improved outcome prediction across data sources through robust parameter tuning. Journal of Classification, 38, 212–31. doi: 10.1007/s00357-020-09368-z
18.
EtagaHO, NdubisiRC and OluebubeNL (2021) Effect of multicollinearity on variable selection in multiple regression. Science Journal of Applied Mathematics and Statistics, 9, 141–53. doi: 10.11648/j.sjams.20210906.12
19.
FahrmeirL, KneibT, LangS and MarxB (2013) Regression – Models, Methods and Applications. 1st edition. New York: Springer. doi: 10.1007/978-3-642-34333-9
20.
FanJ and HuangL-S (2001) Goodness-of-fit tests for parametric regression models. Journal of the American Statistical Association, 96, 640–52. doi: 10.1198/016214501753168316
21.
FriedmanJH (1991) Multivariate adaptive regression splines. The Annals of Statistics, 19(1), 1–67. doi: 10.1214/aos/1176347963
22.
FriedmanJH (2001) Greedy function approximation: A gradient boosting machine. The Annals of Statistics, 29, 1189–1232. doi: 10.1214/aos/1013203451
23.
FriedmanJH, HastieT and TibshiraniR (2010) Regularization paths for generalized linear models via coordinate descent. Journal of Statistical Software, 33, 1–22. doi: 10.18637/jss.v033.i01
24.
GertheissJ and OehrleinF (2011) Testing linearity and relevance of ordinal predictors. Electronic Journal of Statistics, 5, 1935–59. doi: 10.2307/2986270
25.
GülerZO and BakírMA (2019) Performance of methods determining structural break in linear regression models. International Econometric Review, 11, 70–83. doi: 10.33818/ier.450804
HastieT and TibshiraniR (1993) Varying-coefficient models. Journal of the Royal Statistical Society Series B: Statistical Methodology, 55, 757–79. doi: 10.1111/j.2517-6161.1993.tb01939.x
28.
HastieT, TibshiraniR and FriedmanJ (2009) The Elements of Statistical Learning, 2nd edition. New York: Springer. doi: 10.1007/ 978-0-387-84858-7
29.
HeinzeG, AmblerG and BennerA (2023) mfp: Multivariable Fractional Polynomials. R package version 1.5.4. https://CRAN.R-project.org/package=mfp
30.
HofnerB, HothernT, KneibT and SchmidM (2011) A framework for unbiased model selection based on boosting. Journal of Computational and Graphical Statistics, 20, 956–71. doi: 10.1198/jcgs.2011.09220
31.
HolterJC and StallrichJW (2023) Tuning parameter selection for penalized estimation via R. Computational Statistics and Data Analysis, 183. doi: j.csda.2023.107729
32.
InkerLA, SchmidCH, TighiouartH, EckfeldtJH, FeldmanHI, GreeneT, KusekJW, ManziJ, LenteFV, ZhangYL, CoreshJ and LeveyAS (2012) Estimating glomerular filtration rate from serum creatinine and cystatin C. New England Journal of Medicine, 367, 20–9. doi: 10.1056/nejmoa1114248
33.
JankováJ, ShahRD, BühlmannP and SamworthRJ (2020) Goodness-of-fit testing in high dimensional generalized linear models. Journal of the Royal Statistical Society Series B: Statistical Methodology, 82, 773–95. doi: 10.1111/rssb.12371
34.
KneibT, HothornT and TutzG (2009) Variable selection and model choice in geoadditive regression models. Biometrics, 65, 626–34. doi: 10.1111/j.1541-0420.2008.01112.x
35.
Lopez-GiacomanS (2015) Biomarkers in chronic kidney disease, from kidney function to kidney damage. World Journal of Nephrology, 4, 57–73. doi: 10.5527%2Fwjn.v4. i1.57
36.
MayrA, WistubaT, SpellerJ, GudeF and HofnerB (2023) Linear or smooth? enhanced model choice in boosting via deselection of base-learners. Stat Mod, 23. doi: 10.1177/1471082X231170045
37.
MilborrowS (2023) earth: Multivariate Adaptive Regression Splines. R package version 5.3.2. https://CRAN.R-project.org/package=earth.
38.
MillerRG and SiegmundDO (1982) Maximally selected chi square statistics. Biometrics, 38, 1011.
39.
Mula-AbedWAS, RasadiKA and Al-RiyamiD (2012) Estimated glomerular filtration rate (eGFR): A serum creatinine-based test for the detection of chronic kidney disease and its impact on clinical practice. Oman Medical Journal, 27, 108–113. doi: 10.5001/omj.2012.23
40.
MurdochWJ, SinghC, KumbierK, Abbasi-AslR and YuB (2019) Definitions, methods, and applications in interpretable machine learning. Proceedings of the National Academy of Sciences, 116, 22071–80. doi: 10.1073/pnas.1900654116
41.
MillerRG and SiegmundDO (1982) Maximally selected chi square statistics. Biometrics, 38, 1011.
42.
NathanDM, BalkauB, BonoraE, Borch-JohnsenK, BuseJB, ColagiuriS, DavidsonMB, De-FronzoR, GenuthS, HolmanRR, JiL, KirkmanS, KnowlerWC, SchatzD, ShawJ, SobngwiE, SteffesM, VaccaroO, WarehamN, ZinmanB and KahnR (2009) International expert committee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes Care, 32, 1327–34. doi: 10.2337%2Fdc09-9033
43.
NelsonSLP, RamakrishnanV, NietertPJ, KamenDL, RamosPS and WolfBJ (2017) An evaluation of common methods for dichotomization of continuous variables to discriminate disease status. Communications in Statistics: Theory and Methods, 46, 10823–10834. doi: 10.1080/03610926.2016.1248783
44.
RoystonP and AltmanDG (1994) Regression using fractional polynomials of continuous covariates: Parsimonious parametric modelling. Applied Statistics, 43, 429–67.
45.
RoystonP and SauerbreiW (2008) Multivariable Model - Building: A Pragmatic Approach to Regression Analysis based on Fractional Polynomials for Modelling Continuous Variables. Hoboken, NJ: Wiley. doi: 10.1002/9780470770771
46.
SauerbreiW, PerperoglouA, SchmidM, AbrahamowiczM, BecherH, BinderH, DunklerD, HarrellFE, RoystonP and HeinzeG (2020) State of the art in selection of variables and functional forms in multivariable analysis: Outstanding issues. Diagnostic and Prognostic Research, 4. doi: 10.1186/s41512-020-00074-3
47.
SelaRJ and SimonoffJS (2012) RE-EM tree: A data mining approach for longitudinal and clustered data. Machine Learning, 86, 169–207. doi: 10.1007/s10994-011-5258-3
48.
ShahRD and BühlmannP (2018) Goodness-of-fit tests for high dimensional linear models. Journal of the Royal Statistical Society Series B: Statistical Methodology, 80, 113–35. doi: 10.1111/rssb.12234
49.
StroblC, BoulesteixAL and AugustinT (2007) Unbiased split selection for classification trees based on the gini index. Computational Statistics & Data Analysis, 52, 483–501. doi: 10.1016/j.csda.2006.12.030
50.
StroemerA, StaerkC, KleinN, WeinholdL, TitzeS and MayrA (2022) Deselection of base-learners for statistical boosting—with an application to distributional regression. Statistical Methods in Medical Research, 31. doi: 10.1177/0962280221105108
51.
SuX, C-LTsai and WangMC (2008) Tree-structured model diagnostics for linear regression. Machine Learning, 74, 111–31. doi: 10.1007/s10994-008-5080-8
52.
SulaimanMK (2019) Diabetic nephropathy: recent advances in pathophysiology and challenges in dietary management. Diabetology & Metabolic Syndrome, 11. doi: 10.1186/s13098-019-0403-4
53.
TherneauTM and AtkinsonEJ (2023) An Introduction to Recursive Partitioning Using the RPART Routines. Mayo Foundation.
54.
TibshiraniR (1996) Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society Series B: Statistical Methodology, 58, 267–88. doi: 10.1111/j.2517-6161.1996.tb02080.x
55.
TitzeS, SchmidM, KottgenA, BuschM, FloegeJ, WannerC, KronenbergF, EckardtKUfor the GCKD study investigators, EckardtKU, TitzeS, ProkoschHU, BarthleinB, BeckA, GanslandtT, GefellerO, SchmidM, KosterJ, MalzerM, SchlieperG, EitnerF, MeisenS, KehlK, ArweilerE, FloegeJ, SchaeffnerE, Baid-AgrawalS, SchindlerR, TitzeS, HubnerS, DienemannT, HilgersKF, EckardtKU, KottgenA, SchultheissU, WalzG, KielsteinJT, LorenzenJ, HallerH, SommererC, ZeierM, BuschM, PaulK, WolfG, HilgeR, SitterT, KraneV, SchmiedekeD, ToncarS, WannerC, EkiciAB, ReisA, ForerL, SchonherrS, WeissensteinerH, KollertitsB, RaschenbergerJ, KronenbergF, GronwaldW, ZachariasH and PO. (2015) Disease burden and risk profile in referred patients with moderate chronic kidney disease: composition of the German Chronic Kidney Disease (GCKD) cohort. Nephrology Dialysis Transplantation, 30, 441–51. doi: 10.1093/ndt/gfr456
56.
WangL, ZhaoH and FanX (2024) Degrees of freedom: Search cost and self-consistency. https://arxiv.org/abs/2308.13630
57.
WoodSN (2017) Generalized Additive Models: An Introduction with R. 2nd edition. Chapman and Hall/CRC. doi: 10.1201/ 9781315370279
58.
YatesLA, AandahlZ, RichardsSA and BrookBW (2022) Cross validation for model selection: A review with examples from ecology. Ecological Monographs, 93. doi: 10.1002/ecm.1557
59.
YeJ (1998) On measuring anc correcting the effects of data mining and model selection. Journal of the American Statistical Association, 93, 120–31.
60.
ZeileisA, KleiberC, KrämerW and HornikK (2003) Testing and dating of structural changes in practice. Computational Statistics & Data Analysis, 44, 109–23. doi: 10.1016/S0167-9473(03)00030-6
61.
ZeileisA, HothornT and HornikK (2008) Model-based recursive partitioning. Journal of Computational and Graphical Statistics, 17. doi: 10.1198/106186008X319331
62.
ZojaC, XinarisC and MacconiD (2020) Diabetic nephropathy: Novel molecular mechanisms and therapeutic targets. Frontiers in Pharmacology, 11, 586892. doi: 10.3389/fphar.2020.586892
