A Regression Analysis of Measured and Modeled Sonic Booms Across A Lateral Microphone Array

This memorandum demonstrates least absolute shrinkage and selection operator (Lasso) regression and compares results with modeled and measured sonic boom data. The data are from the NASA Carpet Determination In Entirety Measurement Phase I (CarpetDIEM I) test flight campaign with an F/A-18 aircraft. Sonic boom predictions are modeled using PCBoom and compared with ground-based field measurements. Lasso regression identifies the relative contributions of 68 parameters to the prediction variability. A reduced-order model (ROM) can be implemented using a smaller subset of the most significant parameters to improve predictions. The results of these regression analyses can help guide future efforts to improve sonic boom predictions. Overall, this work provides a framework for implementing Lasso regression to identify sources of sonic boom measurement variability and prediction misestimation.