Accurately forecasting tropical cyclone intensity remains both a priority and a challenge. USAF and NOAA Hurricane Hunter aircraft routinely measure environmental characteristics within tropical systems that are not currently assimilated in numerical weather prediction models. In this work, we first collect a new dataset of aircraft data from 199 flights by the USAF 53rd Weather Reconnaissance Squadron into 71 named tropical storms between 2014 and 2019. Second, we apply machine learning techniques to this data to predict tropical cyclone intensity and intensity change. Finally, we analyze these results to identify features that are most predictive of these critical storm properties.