Assessing tick attachments to humans with citizen science data: spatio-temporal mapping in Switzerland from 2015 to 2021 using spatialMaxent

Table 2 Summary of model testing results aggregated by year

Year	\(\hbox {AUC}_{\textrm{ROC}}\)	\(\hbox {AUC}_{\textrm{PRG}}\)	PCC	TSS	CBI	COR	MAE	logloss	RMSE
2015	0.89	0.84	0.88	0.75	0.62	0.58	0.41	0.78	0.53
2016	0.9	0.83	0.83	0.67	0.65	0.59	0.43	0.85	0.56
2017	0.92	0.89	0.89	0.78	0.62	0.68	0.37	0.65	0.49
2018	0.82	0.64	0.81	0.62	0.69	0.55	0.42	0.67	0.5
2019	0.83	0.68	0.82	0.65	0.66	0.56	0.41	0.82	0.53
2020	0.86	0.76	0.82	0.64	0.72	0.55	0.41	0.74	0.52
2021	0.92	0.86	0.87	0.75	0.69	0.59	0.43	0.84	0.56

The table displays median results from model testing across various test folds (see Section "Evaluation") for the years 2015 to 2021. The analysis included the following metrics: area under the receiver operating characteristic curve (\(\hbox {AUC}_{\textrm{ROC}}\)), area under the precision-recall-gain curve (\(\hbox {AUC}_{\textrm{PRG}}\)), percent correctly classified (PCC), true skill statistic (TSS), continuous Boyce index (CBI), Pearson correlation (COR), mean absolute error (MAE), root mean square error (RMSE), and log loss (logloss). Values rounded to two decimal digits

ISSN: 1756-3305