Scandium-doped aluminum nitride with a wurtzite structure has emerged as a promising ferroelectric material due to its exceptional physical and chemical properties and its compatibility with existing processing techniques. However, its high coercive voltage presents a substantial challenge for its potential applications. To effectively reduce this high coercive voltage, it is crucial to comprehensively understand the factors governing polarization reversal processes. Unfortunately, a unified set of pivotal factors has not yet been identified. Herein, machine-learning regression models were developed to predict the uniform polarization reversal barrier ( E ua ) using data sets comprising 41 binary and 113 simple ternary wurtzite materials. Features were extracted based on elemental properties, crystal parameters, mechanical properties, and electronic properties. Calculation of E ua and partial feature extraction were performed using first-principles methods. The results revealed that the average cation-ion potential is the primary intrinsic factor influencing E ua . Additionally, the maximum value of the relative height ratio of cations to anions, cell parameter ratio, and average cation Mendeleev number were found to have secondary impacts. This study addresses gaps in the current understanding of E ua , by considering multiple influencing factors beyond a single material system. It contributes to the systematic evaluation of E ua in wurtzite materials, offering valuable insights not only into uniform polarization reversal processes but also as a reference for future research on more complex processes.