Model Development

The DEA itself does not imposed any boundaries on techniques and environments used to develop models. By wrapping pandas.DataFrame closely most of the common Python data science libraries can be used. To use the dea.Cohort() and dea.Encounter() wrappers to develop a model e.g. in PyTorch or TensorFlow you can use the dea.Encounter.dynamic() member to access the DataFrame directly. The dea.Cohort.to_pandas() method provides a convenient way to convert the cohort to a stacked pandas.DataFrame.