The prediction of stellar rotation periods, derived from Kepler data through machine learning techniques, represents a significant advancement in astrophysics. This study evaluates the Random Forest (RF) model’s effectiveness in predicting rotation periods, highlighting its advantages over traditional methods such as Decision Trees (DT) and Gradient Boosting (GB). We employed multiple algorithms to develop a robust predictive model using corrected Kepler light curve data. Initial period estimates were obtained using Lomb-Scargle (LS) periodograms and Transit Least Squares methods, with the dataset divided into training, validation, and testing sets. Our analysis included nine Kepler stars and an additional set of 32 IDs, revealing that RF achieves lower Root Mean Squared Error (RMSE) values and higher accuracy, averaging approximately 85%. For KIC 7199397, the RF model shows an RMSE of 2.1 d compared to 2.7 d for the DT model. The RF method captures periodic signals in stellar light curves, confirming its robustness in modeling complex dynamics. Its ensemble approach mitigates overfitting and enhances generalization, making it well-suited for the noisy nature of stellar data. The RF method’s versatility extends beyond rotation period predictions, making it applicable to various astrophysical phenomena. Predicted periods are presented with uncertainties, enhancing the reliability of estimates. This study validates the RF model as a powerful tool in astrophysical analysis and suggests that machine learning techniques may further improve our understanding of stellar and planetary systems.