This article uses machine learning tools to analyze mechanical field parameters for their impact on the fatigue strength of a structural component. After selecting the geometry, characteristic dimensions were varied to obtain different values of the theoretical notch shape factor. These values were determined using formulas. Finite element simulations were performed for each set of changed dimensions to obtain mechanical field parameters. After establishing a database of mechanical field parameters, their impact on the high-cycle fatigue strength of structural components was determined. It turned out that an important parameter from a fatigue strength point of view is stress triaxiality. The results indicate that for the selected geometry the model explains as much as 97.84% of the variance of the target variable (shape factor), which is a very good result and encourages further analyses for other geometries. The described machine learning application has not been used in the way presented so far, and a positive result will allow replacing the geometry-dependent formulas used to determine the shape factor with a new unified approach based on the stress triaxiality.