In this paper the authors have implemented stiction detection using machine learning algorithm. Valve stiction is a common nonlinearity in pneumatic control valves that can significantly degrade closed-loop performance, often resulting in oscillations and reduced control accuracy. Early detection is essential to prevent performance deterioration and unplanned downtime. In this work, a supervised machine learning based frameworks are proposed for data-driven detection of valve stiction under diverse operating scenarios, including normal operation, stiction, improper controller tuning, and oscillatory disturbances. Random Forest, Gradient Boosting, and Extreme Gradient Boosting (XGBoost) models are trained to classify the presence or absence of stiction in a closed loop system. The models achieved high accuracy on unseen data and demonstrated robustness to process variations and disturbances. The results indicate that XGBoost is the most reliable and robust algorithm for detecting valve stiction, as it effectively distinguishes stiction events while minimizing false predictions.