Inadequate data and spatial dependence in the observations during geochemical studies are among the disturbing conditions when estimating environmental factors contributing to the local variability in the pollutants of interest. Usually, spatial dependence occurs due to the researcher 's imperfection on the natural scale of occurrence which affects the sampling strategy. As a consequence, observations on the study variable are significantly correlated in space. In this study, the machine learning approach was developed and used to study the environmental factors controlling the local variability in fluoride concentrations in drinking water sources of northern Tanzania within the East African Rift Valley. The approach constituted the use of geographical information systems (GIS) technology, exploratory spatial data analysis (ESDA) methods, and spatial regression modeling at a local level. The environmental variables used to study the local variation in fluoride concentration include topography, tectonic processes, water exchanges between hydrogeological layers during lateral movement, mineralization processes (EC), and water pH. The study was based on 20 local spatial regimes determined using GIS based on water sources density in the four hydrogeological environments. Specifically, the nonparametric (one-way Kruskal-Wallis sum ranks test and Multiple Comparisons Dunn Test), spatial statistics (Global Moran 's I statistic), ordinary least squares (OLS) regression, and spatial lag models were used to quantify the effects of topography, tectonic processes, water exchange between hydrogeological environments and water physiochemical parameters (pH and EC) on the spatial variability of fluoride concentrations in drinking water sources at a local scale. In order of significance, the local spatial variation in fluoride concentration is influenced by the EC, topography, tectonic processes, pH, and water exchange between hydrogeological layers during water movement. The results presented in this paper are crucial for safe water access planning in naturally contaminated aquifer systems.