H2O Frame constructed from multiple parts

Question

My training frame is rather large, so I'd like to import them in a way similar to S3's multipart upload. Is the correct way to do this to manually import_file for all the parts, then call rbind on all of these parts? Or is there a more correct way or built-in of doing this?

Answer 1

the function h2o.import_file can handle import from multiple files on it's own. This works both in Python and R.

Python:

    data = h2o.import_file(["/home/some/path/to/airliens/airline1.csv",
                                "/home/some/path/to/airliens/airline2.csv"])

R:

data = h2o.importFile(c("/home/some/path/to/airliens/airline1.csv",
                                "/home/some/path/to/airliens/airline2.csv"))