Makefile commands can be used to manage the project.
The application is containerized, so there is a Dockerfile
for specifying how
the docker image will be constructed. The python requirements are in the
requirements.txt
file.
Folder structure:
src
: application production codetests
: unit tests folder; no tests at the momentnotebooks
: jupyter notebooks used for machine learning setupconf
: config files for each ML modelscripts
: usefull scripts
The jupyter notebooks can be used as a playground for the modelling part.
Prerequisites:
- jupyterlab
- jupytext
The notebooks are saved as python files, since it is better when keeping track with source control.
There are several commands for managing the notebooks from command line (the commands can be checked directly in the makefile):
# open in browser
make notebooks
# save notebooks as python files
make notebook_save
# load notebooks from py files
make notebook_load