Köppen versus the computer: an objective comparison between the Köppen-Geiger climate classification and a multivariate regression tree

Abstract. A global climate classification is defined using a multivariate regression tree (MRT). The MRT algorithm is automated, which removes the need for a practitioner to manually define the classes; it is hierarchical, which allows a series of nested classes to be defined; and it is rule-based, which allows climate classes to be unambiguously defined and easily interpreted. Climate variables used in the MRT are restricted to those from the Köppen-Geiger climate classification. The result is a hierarchical, rule-based climate classification that can be directly compared against the traditional system. An objective comparison between the two climate classifications at their 5, 13, and 30 class hierarchical levels indicates that both perform well in terms of identifying regions of homogeneous temperature variability, although the MRT still generally outperforms the Köppen-Geiger system. In terms of precipitation discrimination, the Köppen-Geiger classification performs poorly relative to the MRT. The data and algorithm implementation used in this study are freely available. Thus, the MRT climate classification offers instructors and students in the geosciences a simple instrument for exploring modern, computer-based climatological methods.

Download Full-text

Köppen versus the computer: comparing Köppen-Geiger and multivariate regression tree climate classifications in terms of climate homogeneity

Hydrology and Earth System Sciences ◽

10.5194/hess-16-217-2012 ◽

2012 ◽

Vol 16 (1) ◽

pp. 217-229 ◽

Cited By ~ 10

Author(s):

A. J. Cannon

Keyword(s):

Multivariate Regression ◽

Global Climate ◽

Regression Tree ◽

Climate Classification ◽

Traditional System ◽

Multivariate Regression Tree ◽

Modern Computer ◽

Computer Based ◽

Algorithm Implementation ◽

Homogeneous Temperature

Abstract. A global climate classification is defined using a multivariate regression tree (MRT). The MRT algorithm is automated, hierarchical, and rule-based, thus allowing a system of climate classes to be quickly defined and easily interpreted. Climate variables used in the MRT are restricted to those from the Köppen-Geiger classification system. The result is a set of classes that can be directly compared against those from the traditional system. The two climate classifications are compared at their 5, 13, and 30 class hierarchical levels in terms of climate homogeneity. Results indicate that both perform well in terms of identifying regions of homogeneous temperature variability, although the MRT still generally outperforms the Köppen-Geiger system. In terms of precipitation discrimination, the Köppen-Geiger classification performs poorly relative to the MRT. The data and algorithm implementation used in this study are freely available. Thus, the MRT climate classification offers instructors and students in the geosciences a simple instrument for exploring modern, computer-based climatological methods.

Download Full-text

The simultaneous inference of zooplanktivorous fish and macrophyte density from sub-fossil cladoceran assemblages: a multivariate regression tree approach

Freshwater Biology ◽

10.1111/j.1365-2427.2008.02124.x ◽

2010 ◽

Vol 55 (3) ◽

pp. 546-564 ◽

Cited By ~ 73

Author(s):

THOMAS A. DAVIDSON ◽

CARL D. SAYER ◽

MARTIN PERROW ◽

METTE BRAMM ◽

ERIK JEPPESEN

Keyword(s):

Multivariate Regression ◽

Regression Tree ◽

Simultaneous Inference ◽

Zooplanktivorous Fish ◽

Multivariate Regression Tree ◽

Tree Approach

Download Full-text

Using a multivariate regression tree to analyze trade-offs between ecosystem services: Application to the main cropping area in France

The Science of The Total Environment ◽

10.1016/j.scitotenv.2020.142815 ◽

2020 ◽

pp. 142815

Author(s):

Gregory Obiang Ndong ◽

Jean Villerd ◽

Isabelle Cousin ◽

Olivier Therond

Keyword(s):

Ecosystem Services ◽

Multivariate Regression ◽

Regression Tree ◽

Multivariate Regression Tree ◽

Trade Offs

Download Full-text

Fibroblast Growth Factor 23 and α-Klotho Protein Are Associated with Adverse Clinical Outcomes in Non-Dialysis CKD Patients

Kidney & Blood Pressure Research ◽

10.1159/000510351 ◽

2020 ◽

Vol 45 (6) ◽

pp. 900-915

Author(s):

Eleni Manou ◽

Elias Thodis ◽

Georgios Arsos ◽

Ploumis Pasadakis ◽

Stylianos Panagoutsos ◽

...

Keyword(s):

Clinical Outcomes ◽

Primary Endpoint ◽

Multivariate Regression ◽

Fibroblast Growth Factor 23 ◽

Regression Tree ◽

Fibroblast Growth ◽

Multivariate Regression Tree ◽

Regression Tree Analysis ◽

Klotho Protein ◽

Fgf 23

Background: Fibroblast growth factor 23 (FGF-23) and α-Klotho protein appear to have an important role in the pathogenesis of CKD-mineral and bone disorders. The aim of this study was to investigate the association of FGF-23 and α-Klotho levels with adverse clinical outcomes in patients with non-dialysis CKD. Materials and Methods: We conducted a prospective cohort study, enrolling participants with non-dialysis CKD from a single center in Greece. At enrollment, glomerular filtration rate (GFR) was measured (mGFR) and plasma levels of carboxyl terminal FGF-23 (cFGF-23) and soluble α-Klotho (sKlotho) were determined by enzyme-linked immunoassay. Participants were followed for up to 5 years or until the occurrence of the primary endpoint of initiation of renal replacement therapy or death. Multivariate regression tree analysis was used to identify informative baseline parameters in order to categorize participants. Also, using median values of cFGF-23 and sKlotho, participants were categorized into 4 groups, in whom survival was compared using Kaplan-Meier and Cox regression analysis. Results: 128 participants were enrolled with a median mGFR of 41.5 mL/min/1.73 m2 (IQR = 28.2). Baseline mGFR correlated with cFGF-23 and sKlotho (r = −0.54 and r = 0.49, respectively; p < 0.0001 for both). cFGF-23 and sKlotho levels correlated negatively (r = −0.24, p = 0.006). Multivariate regression tree analysis resulted in 3 groups defined by cutoff values of mGFR (60.9 mL/min/1.73 m2) and phosphate (3.7 mg/dL). These groups correlated with CKD stage, cFGF-23, and sKlotho (p < 0.0001 for all). During a median follow-up of 36 months (IQR = 22), 40 (31.2%) participants reached the primary endpoint (31 initiated renal replacement therapy, 9 died). Survival to primary endpoint differed among the 4 groups formed using median values of both biomarkers, with the low FGF-23/high Klotho and high FGF-23/low Klotho having the longest and shortest survival, respectively. High FGF-23/low Klotho group, compared to the opposite one, had a significantly elevated risk of the primary outcome (HR, 6.8; 95% CI, 2.3–19.6; p = 0.0004). Conclusions: In patients with CKD stages 1–5, the combination of higher cFGF-23 and lower sKlotho levels along with mGFR and serum phosphate was associated with adverse clinical outcomes. The utility of combinations of traditional and novel biomarkers to predict outcomes warrants further study.

Download Full-text