Add video showing how to select a range of results in a parallel coordinate plot

[ArturoAmorQ]

Following Guillaume's suggestion in this forum comment, a video after this notebook showing how to select a range of results in a parallel coordinate plot may suffice to complement the explanation on how to use this tool.

This may also improve the scores for Quiz M3.2 Q4 and Q5, which are both below 70%.

[ogrisel]

+1

If you write down the script I can give you a review before the recording. Otherwise we can review a draft video.

[ogrisel]

I sorry you already posted the video on mattermost. I thought it was a new version of the previous one. Let me review it now.

[ogrisel]

About the actual content of the video I find it nice and informative and should be good to address the feedbacks on the forum for this module quiz questions.

[ArturoAmorQ]

If you write down the script I can give you a review before the recording.

import pandas as pd

cv_results = pd.read_csv("../figures/randomized_search_results.csv",
                         index_col=0)

cv_results.head()

def shorten_param(param_name):
    if "__" in param_name:
        return param_name.rsplit("__", 1)[1]
    return param_name

(cv_results.rename(
    shorten_param, axis=1).sort_values("mean_test_score", ascending= False)).head()

import numpy as np
import plotly.express as px

fig = px.parallel_coordinates(
    cv_results.rename(shorten_param, axis=1).apply({
        "learning_rate": np.log10,
        "max_leaf_nodes": np.log2,
        "max_bins": np.log2,
        "min_samples_leaf": np.log10,
        "l2_regularization": np.log10,
        "mean_test_score": lambda x: x}),
    color="mean_test_score",
    color_continuous_scale=px.colors.sequential.Viridis,
)
fig.show()

[ogrisel]

(cv_results.rename(
    shorten_param, axis=1).sort_values("mean_test_score", ascending= False)).head()

can be simplified to:

cv_results.rename(shorten_param, axis=1)

as the goal of this cell is just to demo the renaming function, not to analyze the resulting scores if I am not mistaken

[ArturoAmorQ]

as the goal of this cell is just to demo the renaming function, not to analyze the resulting scores if I am not mistaken

Indeed, you are right!