Air quality monitoring is important for public health, but the characteristics of environmental data, which are multivariate, non-normally distributed, and highly correlated, automatically render conventional statistical control methods ineffective. This study aims to develop an air quality monitoring system in Makassar City using a statistical approach, integrating the Multioutput Least Square Support Vector Regression (MLS-SVR) method and the Multivariate Exponentially Weighted Moving Variance (MEWMV) control chart. This study uses daily data on Particulate Matter 2.5 (PM2.5), Sulfur Dioxide (SO2), and Carbon Monoxide (CO) concentrations from January 2024 to August 2025. The methodology involves using MLS-SVR as a whitening process to model nonlinear relationships and eliminate autocorrelation structures. The resulting independent residuals are then monitored using MEWMV control charts to detect shifts in process variability. The results show that the MLS-SVR model produces high prediction accuracy, as evidenced by an R^2 value of 0.7362, an MSE of 0.2638, and an RMSE of 0.5136. The model also successfully removes significant autocorrelation from the raw data. The MEWMV control chart shows good performance by remaining within control limits despite a significant spike in pollutants during Independence Day celebrations, demonstrating the control Chart's ability to distinguish between predictable structural variations and actual process anomalies. This study concludes that the proposed hybrid method provides a sensitive and effective approach to environmental monitoring. Therefore, this integration is recommended to policymakers as an effective method for early detection of air quality deterioration.