Add metric_per_segment_distribution_plot

[Mr-Geekman]

IMPORTANT: Please do not create a Pull Request without creating an issue first.

Before submitting (must do checklist)

• Did you read the contribution guide?
• Did you update the docs? We use Numpy format for all the methods and classes.
• Did you write any new necessary tests?
• Did you update the CHANGELOG?

Type of Change

• Examples / docs / tutorials / contributors update
• Bug fix (non-breaking change which fixes an issue)
• Improvement (non-breaking change which improves an existing feature)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

Proposed Changes

Look #654.

Related Issue

#654.

Closing issues

Closes #654.

[Mr-Geekman]

Example script:

import matplotlib.pyplot as plt
import pandas as pd

from etna.analysis import metric_per_segment_distribution_plot
from etna.datasets import TSDataset
from etna.metrics import MAE
from etna.metrics import SMAPE
from etna.models import ProphetModel
from etna.pipeline import Pipeline


def main():
    df = pd.read_csv("examples/data/example_dataset.csv", parse_dates=["timestamp"])
    df_wide = TSDataset.to_dataset(df)
    ts = TSDataset(df=df_wide, freq="D")

    pipeline = Pipeline(model=ProphetModel(), horizon=10)
    metrics_df, _, _ = pipeline.backtest(ts=ts, n_folds=3, metrics=[MAE(), SMAPE()])

    metric_per_segment_distribution_plot(
        metrics_df=metrics_df,
        metric_name="MAE",
        plot_type="hist",
        per_fold_aggregation_mode=None,
        seaborn_params=dict(multiple="stack"),
    )
    plt.savefig("plot_hist_no_agg")

    metric_per_segment_distribution_plot(
        metrics_df=metrics_df, metric_name="MAE", plot_type="box", per_fold_aggregation_mode=None
    )
    plt.savefig("plot_box_no_agg")

    metric_per_segment_distribution_plot(
        metrics_df=metrics_df, metric_name="MAE", plot_type="violin", per_fold_aggregation_mode=None
    )
    plt.savefig("plot_violin_no_agg")

    metric_per_segment_distribution_plot(
        metrics_df=metrics_df, metric_name="MAE", plot_type="hist", per_fold_aggregation_mode="mean"
    )
    plt.savefig("plot_hist_mean")

    metric_per_segment_distribution_plot(
        metrics_df=metrics_df, metric_name="MAE", plot_type="box", per_fold_aggregation_mode="mean"
    )
    plt.savefig("plot_box_mean")

    metric_per_segment_distribution_plot(
        metrics_df=metrics_df, metric_name="MAE", plot_type="violin", per_fold_aggregation_mode="mean"
    )
    plt.savefig("plot_violin_mean")

    # case of already aggregated metrics
    pipeline = Pipeline(model=ProphetModel(), horizon=10)
    metrics_df, _, _ = pipeline.backtest(ts=ts, n_folds=3, metrics=[MAE(), SMAPE()], aggregate_metrics=True)

    metric_per_segment_distribution_plot(
        metrics_df=metrics_df, metric_name="MAE", plot_type="hist", per_fold_aggregation_mode=None
    )
    plt.savefig("plot_hist_is_agg")


if __name__ == "__main__":
    main()

plot_hist_no_agg:

plot_box_no_agg:

plot_violin_no_agg:

plot_hist_mean:

plot_box_mean:

plot_violin_mean:

plot_hist_is_agg:

[github-actions]

🚀 Deployed on https://deploy-preview-666--etna-docs.netlify.app

[codecov-commenter]

Codecov Report

Merging #666 (8a6de85) into master (c05a44a) will decrease coverage by 0.38%.
The diff coverage is 33.33%.

@@            Coverage Diff             @@
##           master     #666      +/-   ##
==========================================
- Coverage   84.45%   84.06%   -0.39%     
==========================================
  Files         119      119              
  Lines        6322     6370      +48     
==========================================
+ Hits         5339     5355      +16     
- Misses        983     1015      +32     

Impacted Files	Coverage Δ
etna/analysis/plotters.py	21.92% <31.91%> (+0.97%)	⬆️
etna/analysis/__init__.py	100.00% <100.00%> (ø)

📣 Codecov can now indicate which changes are the most critical in Pull Requests. Learn more

[iKintosh]

👍