Accurately estimating the incidence rate of intimate partner violence (IPV) using the traditional social survey method remains difficult, as victims are often unwilling to report their experiences, which leads to a systematic underestimation. To address this issue, this study applies supervised machine learning models to predict IPV incidence in China using the data from Third Wave Survey on the Social Status of Women in China (TWSSSCW 2010). The combination of the Random Under Sampler-ensemble sampling technique and the Random Forest algorithm yielded the best results. The imputed data suggest higher rates of physical, verbal, and cold violence (7.10%, 13.74%, and 21.35%) than those in the original dataset (4.05%, 11.21%, and 17.95%), underscoring the potential of supervised machine learning to improve measurement accuracy in quantitative social science research.