It's recommended to use the Python client whenever possible (i.e., for deployment, training, or predicting). In addition to providing a more user-friendly experience, the Python client has some performance optimizations to make learning and predicting much faster. In particular, the client has a few advantages.
Deploy scikit-learn models with a single line of code
Asynchronous HTTP requests to make training and predicting more efficient
Batch feature data automatically for online learning algorithms
The client requires Python 3.5 or greater.
$pip install mlrequest
For usage, refer to the other examples in the mlrequest docs and see the github page.