LIN-748-tutorial for reusable compoennt

[mingjerli]

Description

Add a tutorial notebook for reusable component

Fixes # (issue)

LIN-748

Type of change

New tutorial

How Has This Been Tested?

[mingjerli]

Skip the final two tensorflow cells due to

1. need a new irrelevant dependencies
2. take a while to train, don't think it's a good idea to run for every test

[andycui97]

Let's add a table of contents somewhere.

[andycui97]

I think in the introduction here, we should also "set the scene". Why would you need a re-usable component and what does it help with?

[andycui97]

This is a gargantuan number of imports.

Im noticing that some of the imports at the top like

import matplotlib.pyplot as plt
from matplotlib import style
import seaborn as sns

aren't used, can we validate al the imports are being used here again?

[andycui97]

Load Data and Save as Artifact

[andycui97]

Typo: Explore

[andycui97]

maybe here we should say

"replacing the features corresponding to the different bands"

in case the reader does not know that the bands are a feature of galaxies.

[andycui97]

Quick description of why we can't use the re-usable function to do this would be good. Something like "that re-usable function only worked with sklearn supported models, but with keras we need to replace the whole model training piece and thus need Reuse Functions to Load Data and do FE.

[andycui97]

Update this link before we merge. Also something to check ... I wonder why this isn't creating a doc under docs/mkdocs/tutorials like in b25cd57

Maybe try merging or rebasing with main and seeing if its because we havent picked up the latest .github/workflows/sync_notebooks.yml file.

[andycui97]

grammar here is weird, maybe

Used an example developing a machine learning model to demonstrate how to save artifacts as checkpoints and create reusable functions and modules to modify and reuse pieces of the workflow to develop and test different types of models.

[andycui97]

What are the 2 different ways? Let's be explicit here.

[andycui97]

I feel like we're missing one more line here ... something to summarize it all and tie into the Lineapy theme of getting to production quickly like:

As demonstrated in this tutorial, Lineapy re-usable components can streamline a lot of data science and engineering work allowing the modern data team to define and modify data workflows quickly and flexibly. This helps Data Scientists and Data Engineers move these workflows to production and see impact in a faster, more reliable, and safer way.

Please edit the above so that it fits with the rest of the notebook.