A comparison of random forest and its Gini importance with standard chemometric methods for the feature selection and classification of spectral data

Menze, Bjoern; Kelm, B. Michael; Masuch, Ralf; Himmelreich, Uwe; Bachert, Peter; Petrich, Wolfgang; Hamprecht, Fred A.

doi:10.1186/1471-2105-10-213

Cited by 891 publications

(548 citation statements)

References 40 publications

Supporting

Mentioning

529

Contrasting

Unclassified

Order By: Relevance

“…For any random forest classifier, the feature selection can be visualized. Using the "Gini importance" (24) to rank features over all ASTER images of this study (Fig. 2, Bottom Left), we find the information of ASTER channels 3, 5, and 6 to be the most important during training.…”

Section: Mapping Anthrosols At a Large Scalementioning

confidence: 99%

“…The anthrosol probability maps of observations A-C are shown in the right column (black 100%, white 0%; 50% threshold outlined red). The bottom row shows the feature ranking for all ASTER images of the study using the Gini importance (22,24) as measure (boxes represent quartiles; black lines median; green crosses observations A-C). The final probability map-also shown in Fig.…”

Section: Mapping Anthrosols At a Large Scalementioning

confidence: 99%

See 1 more Smart Citation

Mapping patterns of long-term settlement in Northern Mesopotamia at a large scale

Menze

2012

Proc. Natl. Acad. Sci. U.S.A.

127

104

View full text Add to dashboard Cite

The landscapes of the Near East show both the first settlements and the longest trajectories of settlement systems. Mounding is a characteristic property of these settlement sites, resulting from millennia of continuing settlement activity at distinguished places. So far, however, this defining feature of ancient settlements has not received much attention, or even been the subject of systematic evaluation. We propose a remote sensing approach for comprehensively mapping the pattern of human settlement at large scale and establish the largest archaeological record for a landscape in Mesopotamia, mapping about 14,000 settlement sites-spanning eight millennia-at 15-m resolution in a 23,000-km 2 area in northeastern Syria. To map both low-and high-mounded places-the latter of which are often referred to as "tells"-we develop a strategy for detecting anthrosols in time series of multispectral satellite images and measure the volume of settlement sites in a digital elevation model. Using this volume as a proxy to continued occupation, we find a dependency of the long-term attractiveness of a site on local water availability, but also a strong relation to the relevance within a basin-wide exchange network that we can infer from our record and third millennium B.C. intersite routes visible on the ground until recent times. We believe it is possible to establish a nearly comprehensive map of human settlements in the fluvial plains of northern Mesopotamia and beyond, and site volume may be a key quantity to uncover long-term trends in human settlement activity from such a record.A ssessment of the scale and spatial distribution of human communities in past societies has been a key objective for archaeological research over the last 50 y (1), especially in the Near East, where questions of the origins of urbanism, the state, and empires are almost always approached via a regional perspective (2), and most commonly using the methods of archaeological survey. The pioneering surveys of the 20th century focused on the top of the settlement hierarchy, in the form of the largest mounds (e.g., refs. 3 and 4), but it is now appreciated that by overlooking smaller sites such an approach can produce misleading portraits of settlement systems, particularly those of nonurbanized phases (5). The challenge for archaeologists is to maintain coverage extensive enough to discern significant spatial patterning while increasing survey intensity in order to locate these smaller sites. The first generation of survey archaeologists opted for the former half of this classic tradeoff; more recent generations have increasingly adopted the latter intensive approach.Many surveys have met this challenge by employing various remote sensing datasets. Of the three primary physical properties characterizing sedentary Near Eastern sites since the Neolithicdense surface artifact assemblage, moundedness, and anthropogenic sediments-the latter two can be detected in aerial or satellite imagery and can potentially be recorded at large scale. Indeed, nearly a...

show abstract

Section: Mapping Anthrosols At a Large Scalementioning

confidence: 99%

Section: Mapping Anthrosols At a Large Scalementioning

confidence: 99%

Mapping patterns of long-term settlement in Northern Mesopotamia at a large scale

Menze

2012

Proc. Natl. Acad. Sci. U.S.A.

127

104

View full text Add to dashboard Cite

show abstract

“…More applications of random forests can be found in other different fields like quantitative structure-activity relationship modeling [42], nuclear magnetic resonance spectroscopy [31], or clinical decision supports in medicine in general [11].…”

Section: Some Other Related Applicationsmentioning

confidence: 99%

Random Forest for Bioinformatics

2012

Ensemble Machine Learning

512

344

View full text Add to dashboard Cite

“…Other applications include prediction of patient outcome from highdimensional gene expression data [24,66,5] or proteomic mass spectra classification [30,52], where patients are instances and their outcome is the response to be predicted. Another class of applications deals with the prediction of molecule properties based on sequence information, e.g.…”

Section: Rf Applications In Bioinformatics: Some Examplesmentioning

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

585

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

A comparison of random forest and its Gini importance with standard chemometric methods for the feature selection and classification of spectral data

Cited by 891 publications

References 40 publications

Mapping patterns of long-term settlement in Northern Mesopotamia at a large scale

Mapping patterns of long-term settlement in Northern Mesopotamia at a large scale

Random Forest for Bioinformatics

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

Contact Info

Product

Resources

About