Summary
In this chapter, we started by understanding what the drawbacks of POJOs are. Then, we learned that H2O created a counterpart to POJOs called MOJOs, which do not have the same issues that POJOs have. Then, we learned what MOJOs are and the benefits of using them over POJOs. We learned that MOJOs are smaller and faster than POJOs. In H2O’s internal experimentation, it was found that MOJOs performed better when working with large ML models.
After that, we learned how to practically extract ML models trained using AutoML as MOJOs. We understood how MOJOs can be downloaded in Python, R, and H2O Flow. Another benefit that we came across with MOJOs was that there is a special function called PrintMojo that can be used to create graphical pictures of ML models that can be read by humans. This also makes understanding the contents of the ML model easy.
Building on top of this knowledge, we implemented an experiment where we used the h2o-genmodel.jar file, along with the...
