This python3 library for Jupyter Notebook enables to integrate with hkube system:
- Algorithms: add (including image built), list and delete algorithms.
- Pipelines: Create and store pipelines, get stored pipelines, etc.
- Execution: execute pipeline, track execution status by a progress bar, get the results, etc. Algorithm could be built and added to Hkube in 3 methods:
- Function/s: pass your algorithm function/s (only 'start' function is must) - requirements.txt is created automatically.
- Algorithm folder: pass your algorithm code folder - requirements.txt is expected.
- Github: pass algorithm project github url - requirements.txt is expected.
- Download the hkube project hkube_notebook
- Make sure you have python3 and Jupyter Notebook installed (Anaconda is recommended)
- Make sure python3 is in your path
- Update pip3 and install dependecies:
$ python3 -m pip install --upgrade pip
$ python3 -m pip install pipreqs
$ python3 -m pip install flask=0.12.2
$ python3 -m pip install tqdm=4.28.1
- Better work with a virtual env (in that case install required packages also in your virtual env)
- Install the library using the following shell commands (notice that library version is taken from setup.py):
$ cd hkube_notebook
$ python3 setup.py develop
$ # make sure hkube_notebook is installed
$ python3 -m pip list | grep hkube
- Run Jupyter Notebook server, open a new python session, import the library and start work
import hkube_notebook
- Example and test notebook for pipeline build, store, execute and track: hkube_notebook.ipynb
- DS pipeline example notebook, including algorithms build and end-to-end DS pipeline: preprocess + split data + train batch + find best model, train and dump it: DS Example Pipeline.ipynb
- Tested: Python 3.6.8 from Anaconda
See: https://packaging.python.org/tutorials/packaging-projects/
- Install/update tools:
$ python3 -m pip install --user --upgrade setuptools wheel
$ python3 -m pip install --user --upgrade twine
- Create account for python test registry at: https://test.pypi.org/account/register/
- Create distribution and upload it to test registry:
$ # create a 'dist' diectory with whl and gz files:
$ python3 setup.py sdist bdist_wheel
$ # upload to python registry:
$ python3 -m twine upload --repository-url <url> dist/*
$ # for python test registry use: <url>=https://test.pypi.org/legacy/
$ # Create account for python test registry at: https://test.pypi.org/account/register/
- In user python environment:
$ python3 -m pip install --index-url <url> hkube_notebook
$ # for python test registry use: <url>=https://test.pypi.org/simple/
$ # now install missing dependency packages...