The current work aims to improve the sustainability of wire electric discharge machining by predicting the wire breakages. Wire breakages are process interruptions which increase the machining time, energy wastage and material consumption. The study is a novel approach to predict process continuity by binomial classification of machining outcomes using kernel-based naive Bayes algorithm. The two classes are labelled as wire breakages and continuous machining. Training dataset consists of 31 experiments according to central composite design of response surface methodology, and wire breakage instances are recorded as response. The input dataset contains four machining parameters, namely pulse on time, pulse off time, servo voltage and wire feed rate, whereas mean gap voltage variation is derived from in-process data. The trained model was 96.7% accurate in wire breakage predictions. Further, nine confirmation tests were conducted to check model adequacy in real-world situations. The model predicted all instances of wire breakages accurately. The stages of wire wear up to wire rupture were studied by conducting microstructural analysis. © 2021, The Brazilian Society of Mechanical Sciences and Engineering.