Performance of some variable selection methods when multicollinearity is present

Chong, Il-Gyo; Jun, Chi‐Hyuck

doi:10.1016/j.chemolab.2004.12.011

Cited by 1,684 publications

(1,042 citation statements)

References 7 publications

Supporting

Mentioning

1,013

Contrasting

Unclassified

Order By: Relevance

“…The variable importance in the projection (VIP) values in PLS mainly reflected the correlation of the terms to all the responses [29]. It was found that VIP values of acetonitrile, water, methanol, chloroform and ethanol were 1.303, 0.887, 0.602, 0.588 and 0.551, respectively.…”

Section: Selection Of Extraction Solventsmentioning

confidence: 99%

Method for liver tissue metabolic profiling study and its application in type 2 diabetic rats based on ultra performance liquid chromatography–mass spectrometry

Huang

Yin

Wang

et al. 2011

Journal of Chromatography B

View full text Add to dashboard Cite

Section: Selection Of Extraction Solventsmentioning

confidence: 99%

Method for liver tissue metabolic profiling study and its application in type 2 diabetic rats based on ultra performance liquid chromatography–mass spectrometry

Huang

Yin

Wang

et al. 2011

Journal of Chromatography B

View full text Add to dashboard Cite

“…Variables with a VIP value >1.5 were considered important in discriminating between groups (31) and were selected as the most relevant to explain the differences in metabolic profile. While a VIP >1 threshold is generally accepted (32)(33)(34), the cut-off applied in this study is more restrictive, reducing the possibility of obtaining false positive results.…”

Section: Exploratory Data Analysis and Orthogonal Signal Correction Pmentioning

confidence: 99%

Metabolomic Pattern Analysis after Mediterranean Diet Intervention in a Nondiabetic Population: A 1- and 3-Year Follow-up in the PREDIMED Study

et al. 2014

View full text Add to dashboard Cite

The Mediterranean diet (MD) is considered a dietary pattern with beneficial effects on human health. The aim of this study was to assess the effect of an MD on urinary metabolome by comparing subjects at 1 and 3 years of follow-up, after an MD supplemented with either extra-virgin olive oil (MD+EVOO) or nuts (MD+Nuts), to those on advice to follow a control low-fat diet (LFD). Ninety-eight non-diabetic volunteers were evaluated, using metabolomic approaches, corresponding to MD+EVOO (n=41), MD+Nuts (n=27) or LFD (n=30) groups. The 1 H-NMR urinary profiles were examined at baseline, after 1 and 3 years of follow-up. Multivariate data analysis (OSC-PLS-DA and HCA) methods were used to identify the potential biomarker discriminating groups, exhibiting a urinary metabolome separation between MD groups against baseline and LFD. Results revealed that the most prominent hallmarks concerning MD groups were related to the metabolism of carbohydrates (3-hydroxybutyrate, citrate, cis-aconitate), creatine, creatinine, amino acids (proline, Nacetylglutamine, glycine, branched-chain amino acids and derived metabolites), lipids(oleic and suberic acids), and microbial co-metabolites (phenylacetylglutamine, pcresol). Otherwise, hippurate, trimethylamine-N-oxide, histidine and derivates (methylhistidines, carnosine, anserine) and xanthosine were predominant after LFD. The application of NMR-based metabolomics enabled the classification of individuals regarding their dietary pattern and highlights the potential of this approach for evaluating changes in the urinary metabolome at different time points of follow-up in response to specific dietary interventions.2

show abstract

“…principal component regression (PCR), latent variables are chosen in such a way as to provide maximum correlation with a dependent variable; thus, PLS model contains the smallest necessary number of factors. With the increasing number of factors, a PLS regression model converges to that of an ordinary multiple linear regression model (Chong and Jun, 2005). The number of significant principal components for the PLS algorithm is determined using the cross-validation method.…”

Section: Aridity Index (Ai)mentioning

confidence: 99%

Effects of precipitation, temperature and topographic parameters on evergreen vegetation greenery in the Western Ghats, India

Prasad

Badarinath

Eaturu

2008

Intl Journal of Climatology

View full text Add to dashboard Cite

Bi-weekly National Oceanic and Atmospheric Administration-advanced very high-resolution radiometer (NOAA-AVHRR) satellite data covering a fourteen-year time period (1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003) were used to examine spatial patterns in the normalized difference vegetation index (NDVI) and their relationships with environmental variables covering tropical evergreen forests of the Western Ghats, India. NDVI values and corresponding environmental variables were extracted from 23 different forested sites using the NOAA-AVHRR global inventory monitoring and modelling studies (GIMMS) dataset. We specifically used the partial least square (PLS) multivariate regression technique that combines features from principal component analysis and multiple regression to link spatial patterns in NDVI with the environmental variables. PLS regression analysis suggested the two-component model to be the best model, explaining nearly 71% of the variance in the NDVI datasets with relatively good R 2 value of 0.78 and a predicted R 2 value of 0.74. The most important positive predictors for NDVI included Riva's continentality index, precipitation indicators summed over different quarters, average precipitation and elevation. Also, the results from PLS regression clearly suggested that bio-climatic indicators that relied only on precipitation parameters had much more positive influence than indicators that combined both temperature and precipitation together. These results highlight the climatic controls of vegetation vigor in evergreen forests and have implications for monitoring bio-spheric activity, developing prognostic phenology models and deriving land cover maps in the Western Ghats region of India.

show abstract

Performance of some variable selection methods when multicollinearity is present

Cited by 1,684 publications

References 7 publications

Method for liver tissue metabolic profiling study and its application in type 2 diabetic rats based on ultra performance liquid chromatography–mass spectrometry

Method for liver tissue metabolic profiling study and its application in type 2 diabetic rats based on ultra performance liquid chromatography–mass spectrometry

Metabolomic Pattern Analysis after Mediterranean Diet Intervention in a Nondiabetic Population: A 1- and 3-Year Follow-up in the PREDIMED Study

Effects of precipitation, temperature and topographic parameters on evergreen vegetation greenery in the Western Ghats, India

Contact Info

Product

Resources

About