Python

How-to Guides and Useful Links

Setting up virtual environments in the CLI

conda create --name ENVNAME python=3.9.7
conda activate ENVNAME
conda deactivate ENVNAME

Every time you make a new environment, don’t forget to install git! For my CLI (Anaconda Powershell Prompt for Windows 11):

conda install git

pandas

Set default format for displaying numbers; rounding to the nearest number

pd.options.display.float_format = '{:.0f}'.format

IDE stuff

  • Jupyter notebooks with VSCode are normally my go-to, especially if collaborators are comfortable working with Github.
  • If browser-based IDEs aren’t your favorite, Jupyter Ascending seems like a good option for using a text editor of your choice to generate Jupyter notebooks. (note: I haven’t checked this in a while)
  • Colab is another common tool used; it’s not my preference given the workflow for file management, etc. is very different.

PyTorch for Text Classification

Useful resource comparing old torchtext (legacy) to new torchtext: https://lightrun.com/answers/pytorch-text-overview-of-issues-in-torchtext-and-the-plan-for-revamping

File management with pathlib

Use pathlib for relative paths in Python. Docs: https://docs.python.org/3/library/pathlib.html

Stuff to import at the top of the script/NB

import pathlib 
from pathlib import Path

The following lines of code identify a working directory and prints directory name/parent directory.

path = Path.cwd()
print(path)
print(path.parent)

Joining paths: In my workflow, I normally keep the working directory as my code folder and specify the path for storing data collected using relative paths. The following code specifies a path and specifies the file path for the data subdirectory of a project.

code_path = Path.cwd()
data_path = Path(path.parent, 'data')
Back to top