Fig. 2From: SEDE-GPS: socio-economic data enrichment based on GPS informationPerformance of machine learning models predicting microbial lake alpha diversity based on the output of SEDE-GPS. Stars represent the performance of models trained on the respecitve dataset, box plots represent confidence intervals of R2 values gathered from the respective model. Models were trained on the output of SEDE-GPS after feature selection and evaluated using LOOCV (“Methods” section). Only results for the four best-performing models are shown; for the others, see Table 2Back to article page