<p>This study proposes an improved distribution-free multivariate process monitoring framework that uses radial kernel one-class classifier support vector machines (OCC-SVM) to simultaneously monitor ‘location’ and ‘scale’ parameters in a single chart statistic. The proposed unique integrated framework is suitable for monitoring varied sampling scenarios (e.g., individual and rational subgroup of observations). Thus, the work addresses a critical gap in the existing literature, emphasising separate approaches for different sampling scenarios. A novel unsupervised bi-objective heuristic is proposed for near-optimal hyperparameter selection of the OCC-SVM chart, addressing the interaction between kernel bandwidth and penalty factor, an area previously not considered. In the OCC-SVM chart, another challenge is the occurrence of rank-deficient ‘scale’ monitoring matrices when the subgroup size is smaller than the number of quality characteristics. This study addresses this issue by integrating the <i>l</i>₁​-norm graphical lasso (‘glasso’) estimator with an optimal shrinkage parameter, ensuring a well-conditioned (full-rank) matrix for precise ‘scale’ monitoring. Monte-Carlo simulations with various multivariate distributions and real-life case studies validate the effectiveness and robustness of the proposed integrated process monitoring framework compared to existing charts.</p>