[Question] can i adopt the setup if I were to apply it to a jupyter notebook?

[simkimsia]

I saw this setup in the original blogpost.

There's a public jupyter notebook that I am helping the maintainer bring to life as an interactive dashboard. Met him via Chip Huyen's MLOps discord. Same discord where I learn the terms: ibis and dagster.

https://github.com/dgilmore33/NBA-Dashboard

I am thinking of applying your setup (not all of it, just the essential parts) to the jupyter notebook.

I did notice that you mentioned the use of notebook in EDA

I can add from eda import * in a untitled.ipynb for a classic notebook experience in VSCode, which I prefer when plotting or just saving my work.

in the post.

Is it sensible to apply your setup to a jupyter notebook that needs to become an interactive dashboard? Or it's more of a round peg to square hole?

ChatGPT 4 response

This is what I got when I posed the same question to ChatGPT 4 including the link to the post.

Yes, you can apply the setup from the Ibis Analytics project to create an interactive dashboard using Jupyter Notebook. Here’s a general approach to achieve this:

1. Develop Your Analytics Pipeline: Use Ibis for data manipulation and analytics within your Jupyter Notebook, ensuring you can interactively develop and test your data transformations and visualizations.
2. Set Up Streamlit: Streamlit is used for deploying the dashboard. You will need to adapt your Jupyter Notebook code to work within a Streamlit app. This involves organizing your code into functions that Streamlit can use to render the interactive components.
3. Deploy the Streamlit App: Once your Streamlit app is ready, you can deploy it using Streamlit Community Cloud. This involves setting up your app on GitHub and then deploying it from there with a few clicks.
4. Integration and Testing: Ensure that the data flow from Ibis to Streamlit is working correctly, and your interactive components function as expected in the Streamlit environment.

This process involves taking your data manipulation and visualization logic from a Jupyter Notebook and adapting it to a Streamlit app, which can then be deployed as an interactive dashboard.

Let me know if there's any obvious misunderstanding from my question, or any obvious flaws from ChatGPT's response.

[lostmygithubaccount]

Is it sensible to apply your setup to a jupyter notebook that needs to become an interactive dashboard? Or it's more of a round peg to square hole?

I'd say it's sensible! I wouldn't necessarily recommend putting all of the code -- ingestion, ETL with Dagster, and postprocessing -- in the DAG, but once you have the data transformed and loaded into whatever database or table format you want it in, using a notebook for the dashboard/visualization piece makes a lot of sense. there are numerous open source notebook -> dashboard app frameworks, or tools like Hex, that make this pretty easy

you could also put the Dagster logic in a notebook, but personally I'd prefer to keep that in Python scripts. regardless, just importing whatever you need in your notebook and running it is fine. as mentioned, when I'm experimenting I'll open up a notebook, run from eda import * that gives me the con to the database with all the tables + sets up plotly, and go from there

[simkimsia]

Thanks for the response @lostmygithubaccount

I'm going to give it a try

You ok I update you my progress here?

I'll be closing the question

[lostmygithubaccount]

sure thing! feel free to ask questions here or in the Ibis project Zulip (chat)

[simkimsia]

sure thing! feel free to ask questions here or in the Ibis project Zulip (chat)

i joined. Let me know if ok i post my progress here and there

[lostmygithubaccount]

feel free to post in either or both!

[simkimsia]

Crossposting from https://ibis-project.zulipchat.com/#narrow/stream/405263-general/topic/use.20ibis-analytics.20to.20convert.20jupyter.20notebook

In the learning project, I'm doing with a finance analyst, i easily converted the jupyter notebook to be streamlit compatibel script.

i had to swap out the flask/dash related commands. Apparently, not compatible with streamlit but that's easeily done with the help of chat GPT 4. haha :happy:

First time using streamlit and i think it's awesome.

While waiting for feedback from Danny, the finance analyst, I will switch back to replicating Cody's blogpost.

Exciting times!

Here's the latest streamlit-ready python script from the learning project: NBA-Dashboard/NBA.py at main · simkimsia/NBA-Dashboard (github.com)