Neutrino-nucleus modeling uncertainties continue to significantly impact long baseline neutrino oscillation analyses due to complex nuclear interactions and relatively few measurements constraining the vast interaction phase space. Accelerator-based neutrino beams produce large samples of muon (anti) neutrino interactions that have enabled high-statistics, multiply-differential measurements of (anti) neutrino cross sections, but their much smaller rate of electron (anti) neutrino production has limited precision measurements of the electron neutrino counterparts. Furthermore, electron (anti)neutrino interactions have proved difficult to estimate behind significant photonic and wrong-sign backgrounds using model-independent techniques. Here, we present novel data-driven methods for constraining the measured signal and cross-checking the selection efficiency estimation for what will be the first-ever, double-differential measurement of the electron antineutrino charged-current inclusive cross section using data collected by the NOvA experiment from the NuMI beam at Fermilab. We estimate signal events using machine learning and a template-fit to both neutrino and antineutrino enriched data samples; and we perform a data-driven cross check of hadronic effects on signal selection efficiency using a novel simulation technique called Muon-Removed-Electron added (MRE).