Recursive feature elimination with random forest for PTR-MS analysis of agroindustrial products

Granitto, Pablo M.; Furlanello, Cesare; Biasioli, Franco; Gasperi, F.

doi:10.1016/j.chemolab.2006.01.007

Cited by 523 publications

(289 citation statements)

References 22 publications

Supporting

Mentioning

254

Contrasting

Unclassified

Order By: Relevance

“…The variable that resulted in the smallest increase in RMSE was dropped and a new model was created, and new ranks were then calculated for each variable. We continued this process until RMSE reached a minimum [49]. The lowest RMSE occurred when a subset of 12 variables was used to construct Random Forest models.…”

Section: Selection Of Predictor Variablesmentioning

confidence: 99%

Landsat ETM+ and SRTM Data Provide Near Real-Time Monitoring of Chimpanzee (Pan troglodytes) Habitats in Africa

et al. 2016

View full text Add to dashboard Cite

All four chimpanzee sub-species populations are declining due to multiple factors including human-caused habitat loss. Effective conservation efforts are therefore needed to ensure their long-term survival. Habitat suitability models serve as useful tools for conservation planning by depicting relative environmental suitability in geographic space over time. Previous studies mapping chimpanzee habitat suitability have been limited to small regions or coarse spatial and temporal resolutions. Here, we used Random Forests regression to downscale a coarse resolution habitat suitability calibration dataset to estimate habitat suitability over the entire chimpanzee range at 30-m resolution. Our model predicted habitat suitability well with an r 2 of 0.82 (˘0.002) based on 50-fold cross validation where 75% of the data was used for model calibration and 25% for model testing; however, there was considerable variation in the predictive capability among the four sub-species modeled individually. We tested the influence of several variables derived from Landsat Enhanced Thematic Mapper Plus (ETM+) that included metrics of forest canopy and structure for four three-year time periods between 2000 and 2012. Elevation, Landsat ETM+ band 5 and Landsat derived canopy cover were the strongest predictors; highly suitable areas were associated with dense tree canopy cover for all but the Nigeria-Cameroon and Central Chimpanzee sub-species. Because the models were sensitive to such temporally based predictors, our results are the first to highlight the value of integrating continuously updated variables derived from satellite remote sensing into temporally dynamic habitat suitability models to support near real-time monitoring of habitat status and decision support systems.

show abstract

Section: Selection Of Predictor Variablesmentioning

confidence: 99%

Landsat ETM+ and SRTM Data Provide Near Real-Time Monitoring of Chimpanzee (Pan troglodytes) Habitats in Africa

et al. 2016

View full text Add to dashboard Cite

show abstract

“…To reduce the number of features in the final RF model, we started with a model including all input features and made a step-wise reduction by removing each time the least important variable based on the MDA values [50,53,58]. MDA importance values were recalculated at each step.…”

Section: Feature Selectionmentioning

confidence: 99%

Windthrow Detection in European Forests with Very High-Resolution Optical Data

Einzmann

Immitzer

Böck

et al. 2017

Forests

View full text Add to dashboard Cite

With climate change, extreme storms are expected to occur more frequently. These storms can cause severe forest damage, provoking direct and indirect economic losses for forestry. To minimize economic losses, the windthrow areas need to be detected fast to prevent subsequent biotic damage, for example, related to beetle infestations. Remote sensing is an efficient tool with high potential to cost-efficiently map large storm affected regions. Storm Niklas hit South Germany in March 2015 and caused widespread forest cover loss. We present a two-step change detection approach applying commercial very high-resolution optical Earth Observation data to spot forest damage. First, an object-based bi-temporal change analysis is carried out to identify windthrow areas larger than 0.5 ha. For this purpose, a supervised Random Forest classifier is used, including a semi-automatic feature selection procedure; for image segmentation, the large-scale mean shift algorithm was chosen. Input features include spectral characteristics, texture, vegetation indices, layer combinations and spectral transformations. A hybrid-change detection approach at pixel-level subsequently identifies small groups of fallen trees, combining the most important features of the previous processing step with Spectral Angle Mapper and Multivariate Alteration Detection. The methodology was evaluated on two test sites in Bavaria with RapidEye data at 5 m pixel resolution. The results regarding windthrow areas larger than 0.5 ha were validated with reference data from field visits and acquired through orthophoto interpretation. For the two test sites, the novel object-based change detection approach identified over 90% of the windthrow areas (≥0.5 ha). The red edge channel was the most important for windthrow identification. Accuracy levels of the change detection at tree level could not be calculated, as it was not possible to collect field data for single trees, nor was it possible to perform an orthophoto validation. Nevertheless, the plausibility and applicability of the pixel-based approach is demonstrated on a second test site.

show abstract

“…RF could measure the variable importance through "Gini importance" [26] or "permutation importance" [27]. Random forest-recursive feature elimination (RF-RFE) which combines RF with RFE [28] is a recursive backward feature elimination procedure. It begins with all the features.…”

Section: Introductionmentioning

confidence: 99%

Serum metabolomics reveals the deregulation of fatty acids metabolism in hepatocellular carcinoma and chronic liver diseases

et al. 2012

View full text Add to dashboard Cite

Patients with chronic liver diseases (CLD) including chronic hepatitis B and hepatic cirrhosis (CIR) are the major high-risk population of hepatocellular carcinoma (HCC). The differential diagnosis between CLD and HCC is a challenge. This work aims to study the related metabolic deregulations in HCC and CLD to promote the discovery of the differential metabolites for distinguishing the different liver diseases. Serum metabolic profiling analysis from patients with CLD and HCC was performed using a liquid chromatography-mass spectrometry system. The acquired large amount of metabolic information was processed with the random forest-recursive feature elimination method to discover important metabolic changes. It was found that long-chain acylcarnitines accumulated, whereas free carnitine, medium and short-chain acylcarnitines decreased with the severity of the non-malignant liver diseases, accompanied with corresponding alterations of enzyme activities. However, the general changing extent was smaller in HCC than in CIR, possibly due to the special energy-consumption mechanism of tumor cells. These observations may help to understand the mechanism of HCC occurrence and progression on the metabolic level and provide information for the identification of early and differential metabolic markers for HCC.

show abstract

Recursive feature elimination with random forest for PTR-MS analysis of agroindustrial products

Cited by 523 publications

References 22 publications

Landsat ETM+ and SRTM Data Provide Near Real-Time Monitoring of Chimpanzee (Pan troglodytes) Habitats in Africa

Landsat ETM+ and SRTM Data Provide Near Real-Time Monitoring of Chimpanzee (Pan troglodytes) Habitats in Africa

Windthrow Detection in European Forests with Very High-Resolution Optical Data

Serum metabolomics reveals the deregulation of fatty acids metabolism in hepatocellular carcinoma and chronic liver diseases

Contact Info

Product

Resources

About