Stress has become a significant issue in today's society, affecting a person's emotions, thoughts, actions, and communication. Although the human body is designed to cope with stress, prolonged stress can lead to serious health problems such as cancer, cardiovascular disease, depression, and diabetes. However, identifying and tracking a person's stress levels is difficult. Many machine learning methods have been proposed in the literature to address this issue. The primary aim of this study is to develop an accurate and rapid framework for detecting a person's stress levels and estimating their degree of stress. In this proposed study, a machine learning-based approach using an integrated feature selection technique is proposed for human stress detection. The dataset is preprocessed using Label Encoding, and the imbalanced dataset is addressed using SMOTE. The dataset is normalized to ensure consistency. The integrated feature selection technique, Recursive Feature Elimination (RFE), is then applied to select the most crucial features for building the system. Finally, several classification algorithms, such as Decision Tree (DT), Random Forest, Gradient Boosting, AdaBoost, XGBoost, and Extra Tree Classifier, are utilized to identify the stress level of humans. The experiments are conducted on the SWELL-KW dataset.