Flush out content of "Getting Started"

[ablaom]

Also switched out DecisionTree for KNNRegressor, because there is no ambiguity of pkg in that case.

[netlify]

✅ Deploy Preview for mlj ready!

Name	Link
🔨 Latest commit	9b59d56
🔍 Latest deploy log	https://app.netlify.com/sites/mlj/deploys/67494102a54a3f0009dcd56e
😎 Deploy Preview	https://deploy-preview-20--mlj.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site configuration.

[ablaom]

@EssamWisam There is some formatting issues for "Getting Started". The current assumption seems to be all lines in all sections are about he same length.

[EssamWisam]

Could you check now if it's better? I had to create new section for loading data (maybe we can also add some form of preprocessing there?) because it doesn't look that good when I make the height/width variable and it exceeds the height of the bullets section on the left.

I really like the work and have only two recommendations:

• Make the loading data more generic relative to common use by downloading and loading a CSV. For instance:

download("https://raw.githubusercontent.com/JuliaAI/Imbalance.jl/dev/docs/src/examples/bmi.csv", "./")
df = CSV.read("./bmi.csv", DataFrame)

• Add more code in the evaluation section (maybe be borrowing one or two feasutres from here https://juliaai.github.io/MLJ.jl/stable/evaluating_model_performance/)?

Really optional recommendations but it would look nice if the code blocks had relatively a similar number of lines.

[EssamWisam]

I can make the height vary with the amount of code but I may have to do away with the copy button because the code is rendered by a library and it doesn't expose the pre tag to position the copy button relative to it. That was also a reason I made the height constant.

[ablaom]

Thanks @EssamWisam for the prompt response and improvements. Let me get back to you on the suggestions re data. I was trying to avoid adding non-MLJ packages, such as CSV and DataFrames (and Downloads, as Base.download is deprecated).

[ablaom]

We could do something like:

# load Boston dataset from openml.org:
table = OpemML.load(61)

# split data into target (vector, y) and features (table, X):
y, X = unpack(table, ==(:class))

but this is failing because of an unexpected issue I just reported.

[OkonSamuel]

Should this be here?

[EssamWisam]

Thanks @EssamWisam for the prompt response and improvements. Let me get back to you on the suggestions re data. I was trying to avoid adding non-MLJ packages, such as CSV and DataFrames (and Downloads, as Base.download is deprecated).

I think there may be value in using DataFrames just because it also shows others in this starter's code something more representative of the typical MLJ program they want to write.

[EssamWisam]

Hopefully fixed by now. Do you have an idea for something specific to put for the subtitle?

[ablaom]

Okay, I've gone with adding DataFrames and using OpenML.

@EssamWisam Can you review please?

[EssamWisam]

Okay, I've gone with adding DataFrames and using OpenML.

@EssamWisam Can you review please?

Looks good unless you plan to add more evaluation methods.

[ablaom]

Okay @EssamWisam I added an extra evaluation to balance the whitespace.

Can you do me a favour and test the whole "Getting Started" workflow locally. As this may be the only MLJ code an MLJ-curious user runs, it's critical to check it thoroughly.