Stress identification is a complex task and requiring careful consideration of the factors to determine which factor may serve as a key predictor for effective stress prediction. The psychological issues are increasingly prevalent among people and making early detection is both critical and important for timely intervention. This research paper presents a dual approach of stress prediction in humans by integrating the machine learning techniques with statistical methods. It determines the connection between personality traits determined by ten-item personality inventory (TIPI) and perceived stress measured by perceived stress scale (PSS). The methodology encompasses the linear regression, correlation analysis, feature important assessment, Shapley Additive exPLanations (SHAP) analysis, and a set of machine learning algorithms, encompassing support vector regressor, multilayer perceptron in neural networks, random forests, and decision tree regression. Hyperparameter tuning was conducted for neural network to enhance the accuracy and robustness. Various metrics has been calculated to determine the efficacy of models. The findings are presented using the help of diverse visualizations. The Random Forest regressor and tuned neural network emerged as most effective among all machine learning models. Random forest regressor has found to be superior predictive accuracy with 0.9097R2, lowest RMSE (1.33) and MAE (0.36) value which indicates the excellent fit. The tuned neural network provides a robotic predictive framework with 0.8907R2, lowest MAE. Its performance may be enhanced on large datasets and could emerged as the most accurate model of stress prediction. Emotional stability and openness have been identified as crucial predictor of stress using statistical methods involving ANOVA and Kruskal–Wallis test. A significant variation has been observed between the personality characteristics and stress levels by analyzing the p-value and F-statistics. This indicates that individuals with high degree of emotional stability trait is less prone to stress and can easily cope with mental health issues. These outcomes highlight the potential of machine learning models including the Random Forest regressor and the tuned multilayer perceptron regressor to assess the connection between personality traits and perceived stress in humans enabling early detection and timely interventions.