Skip to content

Latest commit

 

History

History
 
 

pluto_notebooks

README.md

BestPractices in Julia

  • Why do this? No particular reason other than I like Julia, it has some advantages (and disadvantages) over python. Plus I find when I code up tutorials or problems on my own I actually see and learn what is happening, so using Julia forces me to do this.
  • Whats different? Julia has similar but different packages that enable easy implementation of data science and machine learning approaches/algorithms. For example rather than pandas to handle data Julia has Database.jl.

All the packages and compatible versions are provided by the manifest.toml and project.toml, so once this repo is cloned, the following commands will set the Julia environment up:

julia> cd("pluto_notebooks")
julia> using Pkg
julia> Pkg.activate(".")
julia> Pkg.instantiate()

This activates the environment and downloads all dependencies and correct versions based on the project.toml. After this setup, a Pluto.jl notebook server can be launched using:

cd pluto_notebooks
julia --project="@." --eval "using Pluto; Pluto.run()"

This should activate the correct Julia environment for the Pluto notebook so all the packages needed packages are available. However, pluto is still a bit rough around and doesn't always find the proper project.toml so the Pkg.activate(".") command is also included in the pluto notebooks. One of the nice things with a Pluto notebook is it is just pure julia code with commented mark-down syntax, so it runs like any julia file. It is also reactive which is good for poking around in a real-time feedback way.

A runtest.jl is provided just for sanity check to ensure, where possible, the julia implementation match with that produced by the original python approach.

Note on CBFV

At the moment the julia version interfaces with the CBFV library via PyCall.jl. This means it is straightforward to use, however, the path to the feature databases in CBFV is determined using getcwd() from pythons os which is well defined when using PyCall.jl. For this reason I had to copy CBFV to this folder and slightly modify it so that now you can provide the file path for the feature database.