A parsimonious model for predicting the NO3−-N concentration in shallow groundwater in intensive agricultural areas using few easily accessible indicators and small datasets based on machine learning

•Groundwater NO3−-N is well predicted using few easily accessible predictors and small datasets.•Electric conductivity, nitrogen fertilization and precipitation are selected as accessible predictors of groundwater NO3−-N.•A parsimonious prediction model with good performance is achieved with a small dataset.•Model balances the accuracy of data on intrinsic factors and the accessibility of data on extrinsic factors.