Model performance was evaluated using the partial receiver operating characteristic (pROC) approach, in addition to the area under the curve (AUC). Partial ROC represents a more suitable indicator of statistical significance and allows a better assessment of the niche model predictive ability [120], considering only omission error and proportional area predicted as suitable, and only over a range of omission error deemed acceptable in light of error characteristics of the input data [136]. AUC measures can be misleading and may reflect model accuracy poorly. It weights omission and commission errors equally, does not give information about the spatial distribution of model errors, and summarizes the entire ROC curve, including regions that frequently are not relevant to practical applications [94, 100]. In a partial ROC test, the statistical significance is determined by bootstrap resampling of 50% of testing data, and probabilities are assessed by direct count of the proportion of bootstrap replicates for which the AUC ratio is ≤1.0 [42]. Occurrence datasets and obtained maps were subjected to over 1000 bootstrap iteration analyses, each based on 50% random points resampling, with replacement, and with an omission error threshold of 1% (p < 0.01). The pROC statistics test was used using the pROC function available in the NicheToolBox package under R system [118].

Note: The content above has been extracted from a research article, so it may not display correctly.



Q&A
Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.



We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.