- Use jupyterlab-unfold for directory tree views:
pip install jupyterlab-unfold - <ctrl,> to toggle settings
- Use Tabulator and Panel for making interactive data explorers: https://panel.holoviz.org/reference/widgets/Tabulator.html
- Use Hvplot Explorer for exploring new data: https://hvplot.holoviz.org/user_guide/Explorer.html - Learn about hvplot explorer here: https://www.youtube.com/watch?v=QbPHVcyAGx8
- Use icecream for logging and printing: https://github.com/gruns/icecream
from icecream import ic
ic.configureOutput(prefix='',outputFunction=print)
ic(len(df[df['success']==False]))
- Use Langchain and OpenAI for exploring dataframes: https://python.langchain.com/docs/integrations/toolkits/pandas
from langchain.agents import create_pandas_dataframe_agent
from langchain.llms import OpenAI
agent = create_pandas_dataframe_agent(OpenAI(temperature=0), df, verbose=False)
command = agent.run("List columns that only have one value. Return a dataframe that removes those columns.")
- Use Python Poetry for dependency management: https://python-poetry.org/
- Use Nbdev for packaging and publish software: https://nbdev.fast.ai/
Thread: https://twitter.com/ygg_anderson/status/1669928219653574656
For #1, careful. I noticed that it's not working with jupyterlab==4.0.3.
I switched to jupyterlab3.6.0 with
poetry add jupyterlab==3.6.0.I find that jupyterlab version 4 is generally buggy.