Fig. 5: Random Forest classification analysis to predict clinical and taxonomic features associated with preterm and full-term delivery. | ISME Communications

Fig. 5: Random Forest classification analysis to predict clinical and taxonomic features associated with preterm and full-term delivery.

From: Longitudinal development of the airway metagenome of preterm very low birth weight infants during the first two years of life

Fig. 5

Microbial taxonomy data, clinical metadata and diversity parameters were included in the model. A Representation of the classification outcome based on mean decrease accuracy. B Representation based on mean decrease Gini. C Overview of the out-of-bag (OOB) estimate of error rates for classifications, which were obtained from 100 times repeated Random Forest and Boruta wrapper application runs with different seeds. D Taxonomical and clinical (other) variables contributing as a predictor for classification as ‘preterm’ or ‘full-term’. The 95% most abundant species were classified as high-abundance taxa.

Back to article page