Variable selection using random forests

Genuer, Robin; Poggi, Jean‐Michel; Tuleau-Malot, Christine

doi:10.1016/j.patrec.2010.03.014

Cited by 2,023 publications

(1,293 citation statements)

References 24 publications

Supporting

Mentioning

1,262

Contrasting

Unclassified

Order By: Relevance

“…In random forest regression, the most widely used index of variable importance is the percentage of increase in mean square error (%IncMSE) following the permutation of a given predictor variable. Higher %IncMSE indicates greater variable importance (Genuer, Poggi, & Tuleau-Malot, 2010). …”

Section: Discussionmentioning

confidence: 99%

Prevalence and predictors of posttraumatic stress disorder and depressive symptoms among burn survivors two years after the 2015 Formosa Fun Coast Water Park explosion in Taiwan

2018

European Journal of Psychotraumatology

View full text Add to dashboard Cite

Background: Posttraumatic stress disorder (PTSD) and depressive symptoms are relatively common in burn survivors. Several previously reported risk factors (e.g. burn severity) have not consistently predicted psychological adjustment post-burn. Empirically-derived risk factors of PTSD from the meta-analysis and theory-derived cognitive variables may be of great predictive value. Objective: This study investigated the prevalence of probable DSM-5 PTSD and major depression (MDD) and the predictors of PTSD and depressive symptoms in burn survivors of the 2015 Formosa Fun Coast Water Park explosion. Three sets of predictors were examined: (a) burn-related variables; (b) empirically-derived risk factors from the meta-analysis; and (c) theory-derived cognitive variables. Method: Participants were 116 burn survivors of the Formosa Fun Coast Water Park explosion. The mean age at the disaster was 22.3 ± 4.2 years; the average total body surface area burned (TBSA) was 49.5%. Results: Of our participants, 12.9 and 20.7% met DSM-5 probable PTSD and MDD two years after the Formosa Fun Coast Water Park explosion. No gender differences were observed. For the prediction of PTSD symptoms post-burn, theory-derived cognitive variables (adjusted R 2 = .562, 95% CI [.423, .638]) performed best and provided significantly better prediction than empirically-derived risk factors from the meta-analysis (adjusted R 2 = .337, 95% CI [.180, .412]) and burn-related variables (adjusted R 2 = .313, 95% CI [.156, .389]). In contrast, the three sets of variables examined provided similar predictions for depressive symptoms post-burn (adjusted R 2 = .267–.295). Random forest regression revealed that theory-derived cognitive variables, particularly negative appraisal of symptoms and maladaptive cognitive coping, were considered the most important predictors of PTSD symptoms post-burn. Conclusion: The prevalence of probable PTSD and MDD were relatively higher in burn survivors. Theory-derived cognitive variables substantially improve predictions for PTSD symptoms post-burn.

show abstract

Section: Discussionmentioning

confidence: 99%

Prevalence and predictors of posttraumatic stress disorder and depressive symptoms among burn survivors two years after the 2015 Formosa Fun Coast Water Park explosion in Taiwan

2018

European Journal of Psychotraumatology

View full text Add to dashboard Cite

show abstract

“…The optimal subset is then the subset yielding the smallest error frequency [24] or the smallest area under the curve [14]. An alternative variable selection approach based on a nested collection of random forests is described in Genuer et al [29]. Again, it needs to be emphasized that the resulting model with selected variables needs to be externally validated.…”

Section: Random Forests and Variable Selectionmentioning

confidence: 99%

Overview of random forest methodology and practical guidance with emphasis on computational biology and bioinformatics

Boulesteix

Janitza

Kruppa

et al. 2012

WIREs Data Min & Knowl

632

455

View full text Add to dashboard Cite

The Random Forest (RF) algorithm by Leo Breiman has become a standard data analysis tool in bioinformatics. It has shown excellent performance in settings where the number of variables is much larger than the number of observations, can cope with complex interaction structures as well as highly correlated variables and returns measures of variable importance. This paper synthesizes ten years of RF development with emphasis on applications to bioinformatics and computational biology. Special attention is given to practical aspects such as the selection of parameters, available RF implementations, and important pitfalls and biases of RF and its variable importance measures (VIMs). The paper surveys recent developments of the methodology relevant to bioinformatics as well as some representative examples of RF applications in this context and possible directions for future research.

show abstract

“…An RF is essentially an ensemble method that constructs a multitude of decision trees (each of them trained with different subsets of features and examples), and yields the mean prediction of the individual trees [51]. RFs' classification and regression have been applied in different areas of concern in forest ecology, such as modelling the gradient of coniferous species [52], the occurrence of fire in Mediterranean regions [53], the classification of species or land cover type [54,55], and the analysis of the relative importance of the proposed drivers [55] or the selection of drivers [54,56,57]. The selection of RFs in our study is not incidental, and we capitalize on several useful properties.…”

Section: Random Forests Regressionmentioning

confidence: 99%

Global Estimation of Biophysical Variables from Google Earth Engine Platform

et al. 2018

View full text Add to dashboard Cite

This paper proposes a processing chain for the derivation of global Leaf Area Index (LAI), Fraction of Absorbed Photosynthetically Active Radiation (FAPAR), Fraction Vegetation Cover (FVC), and Canopy water content (CWC) maps from 15-years of MODIS data exploiting the capabilities of the Google Earth Engine (GEE) cloud platform. The retrieval chain is based on a hybrid method inverting the PROSAIL radiative transfer model (RTM) with Random forests (RF) regression. A major feature of this work is the implementation of a retrieval chain exploiting the GEE capabilities using global and climate data records (CDR) of both MODIS surface reflectance and LAI/FAPAR datasets allowing the global estimation of biophysical variables at unprecedented timeliness. We combine a massive global compilation of leaf trait measurements (TRY), which is the baseline for more realistic leaf parametrization for the considered RTM, with large amounts of remote sensing data ingested by GEE. Moreover, the proposed retrieval chain includes the estimation of both FVC and CWC, which are not operationally produced for the MODIS sensor. The derived global estimates are validated over the BELMANIP2.1 sites network by means of an inter-comparison with the MODIS LAI/FAPAR product available in GEE. Overall, the retrieval chain exhibits great consistency with the reference MODIS product (R 2 = 0.87, RMSE = 0.54 m 2 /m 2 and ME = 0.03 m 2 /m 2 in the case of LAI, and R 2 = 0.92, RMSE = 0.09 and ME = 0.05 in the case of FAPAR). The analysis of the results by land cover type shows the lowest correlations between our retrievals and the MODIS reference estimates (R 2 = 0.42 and R 2 = 0.41 for LAI and FAPAR, respectively) for evergreen broadleaf forests. These discrepancies could be attributed mainly to different product definitions according to the literature. The provided results proof that GEE is a suitable high performance processing tool for global biophysical variable retrieval for a wide range of applications.

show abstract

Variable selection using random forests

Cited by 2,023 publications

References 24 publications

Prevalence and predictors of posttraumatic stress disorder and depressive symptoms among burn survivors two years after the 2015 Formosa Fun Coast Water Park explosion in Taiwan

Prevalence and predictors of posttraumatic stress disorder and depressive symptoms among burn survivors two years after the 2015 Formosa Fun Coast Water Park explosion in Taiwan

Overview of random forest methodology and practical guidance with emphasis on computational biology and bioinformatics

Global Estimation of Biophysical Variables from Google Earth Engine Platform

Contact Info

Product

Resources

About