One of the ways to reduce the aerodynamic drag is by improving the rear wake of the vehicle. To achieve this, accurate measurements of flow velocity and pressure at the rear of the vehicle are essential. In the Hyundai Aero-Acoustic Wind Tunnel (HAWT), a wake measurement system with cobra probe arrays has been installed to measure and analyze the vehicle wake. However, the cobra probe can only properly detect the flows above 10m/s; thus, it cannot create an accurate wake contour within the range of -10m/s to 40m/s. In this paper, a Physics-Informed Neural Network (PINN) is applied to reconstruct the complete vehicle wake from this sparse data. The PINN model fills in the flow where velocities are below a certain threshold, allowing for a precise calculation of micro drag, a quantitative method for vehicle wake analysis. The generic aerodynamic model DrivAer, simulated via CFD, is used to validate the predictive accuracy of the PINN. As a result, the difference in aerodynamic drag coefficient between the ground truth and the PINN prediction is under 1 count (ΔC_D < 0.001). Furthermore, the validated model was successfully applied to live wind tunnel data from the Hyundai IONIQ 5, enabling a quantitative diagnosis that guided a significant drag reduction. The application of PINN is expected to establish a more precise and practical technique for vehicle wake analysis.