The current research focuses on the influence of the geotechnical parameters such as depth, groundwater head, excess Pore pressure, pore pressure ratio, vertical displacement, and pore water pressure on the suction efficiency of different construction sites. In this paper, a hybrid of FEA and machine learning through a Random Forest algorithm provides analysis and prediction for the aforementioned variables. FEA was carried out for soil displacement and pore pressure in several excavation scenarios, while FEA outputs were fed into the Random Forest model to develop variable importance and predict suction efficiency. Results show that the pore water pressure is the feature that most dominates the dependence of suction efficiency; its importance value is essentially incommensurable with that of the rest of the other variables. Among these, more influential factors are groundwater head and excess pore pressure, while vertically to the study, combining FEA and Random Forest helps us understand how geotechnical variables affect the efficiency of suction in a more detailed way. It is also a useful tool for designing and managing groundwater drainage systems in construction projects. Displacement and depth have a lower influence. The Random Forest model presented a high degree of accuracy with strong performance metrics, confirming that the machine learning and numerical simulation combination may yield more accurate and comprehensive predictions than traditional methods. The study concludes that integration of the FEA and Random Forest provides an in-depth understanding of the quantitative relationships of geotechnical variables with the efficiency of suction, besides being a useful means for the design and management of the groundwater drainage system in construction projects. It holds great scientific relevance and practical applications for projects involving soil treatment and groundwater extraction systems.