[hotfix] Fix the bug of matplotlib's plot_rank function

[c-bata]

Motivation

Fix #5115

Description of the changes

Fix the bug of matplotlib's plot_rank() function.

import optuna
from optuna.visualization.matplotlib import plot_rank
import matplotlib.pyplot as plt


def objective(trial):
    x = trial.suggest_uniform('x', -10, 10)
    y = trial.suggest_uniform('y', -10, 10)
    return (x - 2) ** 2 + y

def main():
    study = optuna.create_study()
    study.optimize(objective, n_trials=100)
    plot_rank(study)
    plt.savefig("./foo.png", bbox_inches="tight", dpi=100)

if __name__ == '__main__':
    main()

[codecov]

Codecov Report

Attention: 4 lines in your changes are missing coverage. Please review.

Comparison is base (bc4ee3e) 89.43% compared to head (d093d2f) 89.41%.
Report is 1 commits behind head on master.

Files	Patch %	Lines
optuna/visualization/_rank.py	71.42%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #5133      +/-   ##
==========================================
- Coverage   89.43%   89.41%   -0.02%     
==========================================
  Files         205      205              
  Lines       15151    15160       +9     
==========================================
+ Hits        13550    13556       +6     
- Misses       1601     1604       +3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[not522]

@eukaryo @nabenabe0928 Could you review this PR?

[c-bata]

The windows CI will be resolved by #5135 .

[nabenabe0928]

Hi thank you for the quick fix!

I checked the color map array for matplotlib and plotly just in case with the following code:

import plotly
import matplotlib.pyplot as plt
import numpy as np


def get_color_scale_in_plotly(color_idxs: np.ndarray) -> np.ndarray:
    colormap = "RdYlBu_r"
    # sample_colorscale requires plotly >= 5.0.0.
    labeled_colors = plotly.colors.sample_colorscale(colormap, color_idxs)
    scaled_rgb_colors = np.array([plotly.colors.unlabel_rgb(cl) for cl in labeled_colors])
    return scaled_rgb_colors


def get_color_scale_in_matplotlib(color_idxs: np.ndarray) -> np.ndarray:
    colormap = "RdYlBu_r"
    cmap = plt.get_cmap(colormap)
    colors = cmap(color_idxs)[:, :3]
    rgb_colors = np.asarray(colors * 256, dtype=int)
    return rgb_colors


color_idxs = np.linspace(0, 1, 101)
cmap_plotly = get_color_scale_in_plotly(color_idxs)
cmap_matplotlib = get_color_scale_in_matplotlib(color_idxs)

Although they are not perfectly identical, I suppose the difference does not really matter from the user side, so I will leave it intact!

[c-bata]

Thank you for your review. PTAL.

[nabenabe0928]

Thank you for the update! LGTM!

[contramundum53]

The code looks good to me. Let me check the behavior.

[contramundum53]

The code works perfect. LGTM!

import optuna
study = optuna.create_study()

def objective(trial):
    x = trial.suggest_uniform('x', -10, 10)
    y = trial.suggest_uniform('y', -10, 10)
    return (x - 2) ** 2 + (y + 3) ** 2

study.optimize(objective, n_trials=100)

import optuna.visualization.matplotlib
optuna.visualization.matplotlib.plot_rank(study)

[c-bata]

@contramundum53 @nabenabe0928 I've just noticed a minor bug where the RGB values could potentially reach 256, though they should be within the range [0, 255].

Could you please take a look at the following change?
d093d2f

[nabenabe0928]

I checked the change!

[contramundum53]

LGTM!