Understanding what a MOJO is

MOJOs are counterparts to H2O model POJOs and technically work in the same way. H2O can build and extract models trained in the form of MOJOs, and you can use the extracted MOJOs to deploy and make predictions on inbound data.

So, what makes MOJOs different from POJOs?

POJOs have certain drawbacks that make them slightly less than ideal to use in a production environment, as follows:

• POJOs are not supported for source files larger than 1 GB, so any models with a size larger than 1 GB cannot be compiled to POJOs.
• POJOs do not support stacked ensemble models or Word2Vec models.

MOJOs, on the other hand, have the following additional benefits:

• MOJOs have no size restrictions
• MOJOs solve the large size issue by removing the ML tree and using a generic tree walking algorithm to navigate the model computationally
• MOJOs are smaller in size and faster than POJOs
• MOJOs support all types of models trained using H2O AutoML...