In this paper we present a systematic data-driven parameter correction and estimation process consisting of outlier detection and removal, relevant input parameter selection, advanced statistical and empirical correlation, and prediction fusion to reduce variance in relevant engine parameter estimates. We model engine parameter deviations from nominal, and show that these methods can result in significant reductions in bias and variance modeling errors. Reducing the error variance increases the signal-to-noise ratio, thereby increasing the reliability and speed of fault-detection algorithms. The overall objective function is to reduce the measurement variances without masking faults. Key parameters modeled include fuel flow, rotor speed(s), and measured temperatures.