Curated list of links related to coding in economics. We will try to list links to class content, notebooks, replication codes, libraries and data sources. We are just starting this so bare with us and please send us suggestions!
- RISE Jupyter slideshow extension
- Google Collab online Jupyter notebooks with GPU support
- Atom text editor
- Interactive coding with hydrogen
- Julia support with IJulia and language-julia
- R support with IRkernel and language-r
- VS code open source text editor by Microsoft.
- I find that using the notebook in the browser is better than the notebook inside VScode
- Jupytext seamless sync between text based files and notebook format (json)
This is the first required piece. The following resources help with portability of your code.
- conda is a the goto package manager for data-science. Ihttps://vifm.info/t supports all platform, and provides binaries so you don't need tool chains installed. It also doesn't require admin right on the machine. It supports all languages like julia, R, go, rust and of course python
- conda-build allows to port a conda environment prepared on one machine to another.
- mamba is a drop-in replacement for conda that is faster.
- Nix is a manage it all manager. It requires admin rights.
Container approach: think of these as thin virtual machine. This is the easier to port as long as the host machine has the ability to run containers.
- docker is an industry standard for portable code.
- singularity is probably where academia will converge.
Some language specific package and environment managers:
- julia natively supports environment. It's really great. It can be hard however to relocate an environment.
- pdm is a recent python enivronment manager, compared to the king of the gender
npm - peotry is a very friendly python environment manager. Makes is easy to upload package to pip
- Julia
- DataFrames.jl main pacakge to handle dataframes in Julia
- DataFramesMeta.jl handy additional functions
- Chain.jl piping, recommmned to use with the above
- Python
- Pandas de facto dataframe for python
- Method chaining to improve readability
- Modin dropin replacement for large datasets
- R
- data.table fast dataframe for R
- Pytorch
- Installation with or without GPU support
- Quickstart to learn the basics
- JAX
- Keras
- Flux.l in julia
- scikit-learn in Python
- mlr3 in R
- GitHub Pages to host a private website
- Jekyll for building pages from HTML/CSS
- reveal.js is the most mature but also older.
- mdx-deck based on react and MDX (markdown)
- spectacle also react
- slidev is based on vue.js and vite.js. This is the best html framework to write academic presentation. It supporst math, annotations, code highlights and the list goes on and on. Because it is built on vite, the slides get updated almost instantly whenever you change the source. It's amazing!
- Make, the classic
- nextflow
- doit
- snakemake
- luigi, airflow and prefect are great but not clearly adapted to scientific work