Classification

The metaflow classification example is the same as the normal classification example. The only difference is that it users metaflow as the ML pipeline tool. You can follow the instructions in the classification example, with the following two modifications:

1. When installing libraries for the example, be sure to install lolpop[metaflow] in addition to the other extras.

2. When running the example you'll want to run it from the same directory, i.e. lolpop_petfinder_example, but run it via this command:

python3 metaflow/run.py 

The rest of it should look fairly identical.

Note

When using the metaflow pipelines, when you're logging to stdout, the logging will be be a bit redundant, as it currently applies both the metaflow and lolpop formats. In the future we may try to make this less noisy -- a quick fix if this annoys you is to modify the StdOutLogger class to just not include the redundant pieces.