The goal of our study is to develop machine learning models to predict the hospital length of stay from open healthcare data. The length of stay is an important metric of the performance of healthcare systems. It can be used to drive cost efficiencies and allocate resources. We analyzed de-identified patient data from New York State SPARCS (statewide planning and research cooperative system), consisting of 2.3 million patient records. We investigated multiple model categories consisting of regression, decision trees, random forests, and XGBoost. The most important features we identified were the diagnostic related group and the severity of illness code. The best regression model was XGBoost, with an R2 value of 0.44.