The safety of property and people against risks and electrical faults is essentially ensured by an earthing system. The essential element of the latter is the electrical resistivity of the soil. This is influenced by several factors. In this work, we used the georeference coordinates of the point (A), state of nature the day before the measurement (B), state of nature on the day of the measurement (C), and the ambient temperature at the measurement point (D). 9513 in-situ measurement samples on selected sites in Lomé constitute the database. Genetic algorithms and fuzzy inference systems made it possible to create models taking into account the bibliographic review. Evaluation criteria such as: RMSE, RRMSE, MAPE and R² are used to evaluate these models. The results give RMSE = 16.20%, RRMSE = 10.94%, MAPE = 9.27%, R² = 97.93% for Genetic Algorithms with the BCD configuration and for fuzzy inference systems, we have RMSE = 71.48, RRMSE = 48.29%, MAPE =35.29%, R² =61.53% by ACD configuration. We conclude that Genetic Algorithms give a very good result given the value of its correlation coefficient with the BCD combination, which justifies that the parameters used are well suited to predicting the electrical resistivity of the soil in the area considered.