The hidden nature of labor trafficking makes it difficult to detect labor violations relating to agricultural workers on H-2A visas who are vulnerable to exploitation. This study helps labor inspectors prioritize which worksites to inspect based on publicly available data regarding employer wage and hour violations and H-2A certification applications. A Natural language preprocessing technique is used to match employer’s names which are recorded with typos, spelling errors, and abbreviations, presenting a challenge to merge the visa and labor violations datasets. We analyze the distribution of visa applications for agricultural workers and violations and explore factors that affect the number of labor violations. The purpose of this study is to identify whether there is a correlation between H2A employers and labor violations based on features such as the number of worksites, agricultural sector, number of requested visa workers, and population changes. The method relies on building different predictive models on the training set of visa applications to train the models and then to detect labor violations on the remaining visa application.