The rise of advanced manufacturing in the polymer industry brings with it demands for new sensors and controllers as well as greater computer power. The introduction of its associated technologies has necessitated a steep transition from simpler univariate data collection to managing large databases of multivariate process descriptors. This creates a challenge on the statistical data analysis we perform with these new techniques in order to demonstrate their full potential benefits. In this study, multivariate statistical analysis was used to interpret surface imaging and complex ultrasonic spectroscopic data from rotomolding process. Both densification and thermo-oxidative degradation were monitoring using the proposed nondestructive methods and validated with impact and rheology tests. Optimum operating conditions were identified through multivariate empirical models using data from historical process batches and an evolutionary algorithm. Results from this study demonstrate the potential combination of new sensors and 'big data' statistical analysis to improve reliability and flexibility of polymer processing. The authors believe that dissemination and inspiration of similar works can bring clarity on how to practically implement this new framework of advanced manufacturing.