Random Forest Regression Based on Partial Least Squares Connect Partial Least Squares and Random Forest

Hao, Zhu-lin; Du, Jianqiang; Nie, Bin; Yu, Fanglin; Yu, Riyue; Xiong, Wangping

doi:10.2991/icaita-16.2016.48

Cited by 4 publications

(4 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The relative variable importance was determined by the Partial Least Squares (PLS) based on the weighted sums of the absolute regression coefficients using the importance function in "Random Forest" packages. PLS is an analytical method applicable when the data include more predictions than observations, and it is used to estimate the importance of a predictor variable in a model regression [55][56][57]. In RF, the most commonly used method to analyze the variable importance is the mean decrease in Gini, where the importance value is expressed as a percentage; this corresponds to the mean variable total decrease in node impurity, weighted by the proportion of samples reaching that node in each decision tree in RF, i.e., normalization of importance values [49].…”

Section: Discussionmentioning

confidence: 99%

Influence of Environmental Factors on Forest Understorey Species in Northern Mexico

Maciel-Nájera

González‐Elizondo

Hernández-Díaz

et al. 2021

Forests

View full text Add to dashboard Cite

Background: Understorey plants are key to maintaining forest structure and functioning. They protect the soil, improve its structure and fertility, reduce water run-off and sustain the below-ground biota, amongst other ecological services. However, little is known about the environmental conditions that regulate the occurrence of these plants. This study focuses on determining how canopy cover influences the occurrence of understorey species and identifying the most important soil properties that affect these species. The study area was a pine-oak forest in the Sierra Madre Occidental, an important source of ecological services for northwestern Mexico. Methods: To assess the conditions influencing the presence of herbaceous and shrub species, 25 soil variables were examined in relation to the species occurring in forest gaps and under the canopy. Sampling was conducted in five plots, each of 100 × 100 m. In each plot, 4 subplots, each of 20 × 20 m, were each subdivided in a grid of 2 × 2 m units, in which the presence-absence of herbaceous and shrub species was recorded (2000 units in total). Soil samples were extracted for analysis from the central point in each subplot. Data were analyzed using a Binomial Logistic Model (BLM) and Random Forest (RF) classification. Results: Understorey species were more strongly affected by soil variables than by their location in gaps or below canopy. The concentrations of Ca, P, K, Fe, Na, C, Zn, Mn, nitrates, organic matter, sand, silt, and percentage water saturation were statistically significantly associated with the presence of some plant species, whilst no significant differences were found in regard to preference for gaps or canopy, although several species were more frequent in open areas. Conclusions: Given the importance of the understorey cover in forest system functioning, we propose that understorey should be considered in integrated management and conservation practices for the temperate forests of northern Mexico.

show abstract

Section: Discussionmentioning

confidence: 99%

Influence of Environmental Factors on Forest Understorey Species in Northern Mexico

Maciel-Nájera

González‐Elizondo

Hernández-Díaz

et al. 2021

Forests

View full text Add to dashboard Cite

show abstract

“…As a nonparametric multivariate statistical analysis method, partial least squares (PLS) provides a regression modeling method for multidependent variables to multi-independent variables in order to effectively solve the problem of multicollinearity [14, 15] when there is a high correlation between independent variables. Based on preponderance, some researchers have proposed a series of improved models.…”

Section: Related Workmentioning

confidence: 99%

Feature Selection Method Based on Partial Least Squares and Analysis of Traditional Chinese Medicine Data

Huang

Nie

et al. 2019

Computational and Mathematical Methods in Medicine

Self Cite

View full text Add to dashboard Cite

The partial least squares method has many advantages in multivariable linear regression, but it does not include the function of feature selection. This method cannot screen for the best feature subset (referred to in this study as the “Gold Standard”) or optimize the model, although contrarily using the L1 norm can achieve the sparse representation of parameters, leading to feature selection. In this study, a feature selection method based on partial least squares is proposed. In the new method, exploiting partial least squares allows extraction of the latent variables required for performing multivariable linear regression, and this method applies the L1 regular term constraint to the sum of the absolute values of the regression coefficients. This technique is then combined with the coordinate descent method to perform multiple iterations to select a better feature subset. Analyzing traditional Chinese medicine data and University of California, Irvine (UCI), datasets with the model, the experimental results show that the feature selection method based on partial least squares exhibits preferable adaptability for traditional Chinese medicine data and UCI datasets.

show abstract

“…The idea has been tried in several papers. Reddy et al [26] and Yeh et al [27] sequentially fit PLS regression and regression tree, whereas Hao et al [28] combined PLS regression and regression tree, and additionally combined PLS regression and random forests. All predictors were used for splitting and fitting in the latter, and only univariate response variables were considered in the studies.…”

Section: Introductionmentioning

confidence: 99%

Partial Least Squares Regression Trees for Multivariate Response Data With Multicollinear Predictors

Yu,

Lee,

Cho

2024

IEEE Access

View full text Add to dashboard Cite

Some problems arise in analyzing massive complex data consisting of multivariate response variables and a large number of multicollinear predictor variables, especially when the sample sizes compared to the number of predictors are small. Rather than ordinary linear regression modeling approaches, latent variable regression modeling approaches such as partial least squares regression can be used to capture the relationship between the response and predictor variables for such cases. However, for complex nonlinear relationships between the predictor and the response variable, the performance of inference and prediction using regression modeling approaches can be deflated. Regression trees can capture such complex relationships. Thus, we develop a partial least squares tree modeling algorithm that detects complex relationships and makes precise predictions by integrating the merits of partial least squares and regression trees. It is shown that it has better predictive performance than other methods through simulation and it is demonstrated that it generates interpretable predictive models with real data of usedcar and orthognathic surgrey.

show abstract

Random Forest Regression Based on Partial Least Squares Connect Partial Least Squares and Random Forest

Cited by 4 publications

References 4 publications

Influence of Environmental Factors on Forest Understorey Species in Northern Mexico

Influence of Environmental Factors on Forest Understorey Species in Northern Mexico

Feature Selection Method Based on Partial Least Squares and Analysis of Traditional Chinese Medicine Data

Partial Least Squares Regression Trees for Multivariate Response Data With Multicollinear Predictors

Contact Info

Product

Resources

About