MetricFrame visualizations

import pandas as pd
from fairlearn.datasets import fetch_diabetes_hospital
from sklearn.metrics import accuracy_score, precision_score
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier

from fairlearn.metrics import (
    MetricFrame,
    count,
    false_negative_rate,
    false_positive_rate,
    selection_rate,
)

data = fetch_diabetes_hospital(as_frame=True)
X = data.data
X.drop(columns=["readmitted", "readmit_binary"], inplace=True)
y_true = data.target
X_ohe = pd.get_dummies(X)
race = X['race']

X_train, X_test, y_train, y_test, \
    A_train, A_test = train_test_split(X_ohe, y_true, race, random_state=123)

classifier = DecisionTreeClassifier(min_samples_leaf=10, max_depth=4)
classifier.fit(X_train, y_train)
y_pred = classifier.predict(X_test)

# Analyze metrics using MetricFrame
metrics = {
    "accuracy": accuracy_score,
    "precision": precision_score,
    "false positive rate": false_positive_rate,
    "false negative rate": false_negative_rate,
    "selection rate": selection_rate,
    "count": count,
}
metric_frame = MetricFrame(
    metrics=metrics, y_true=y_test, y_pred=y_pred, sensitive_features=A_test
)
metric_frame.by_group.plot.bar(
    subplots=True,
    layout=[3, 3],
    legend=False,
    figsize=[12, 8],
    title="Show all metrics",
)

# Customize plots with ylim
metric_frame.by_group.plot(
    kind="bar",
    ylim=[0, 1],
    subplots=True,
    layout=[3, 3],
    legend=False,
    figsize=[12, 8],
    title="Show all metrics with assigned y-axis range",
)

# Customize plots with colormap
metric_frame.by_group.plot(
    kind="bar",
    subplots=True,
    layout=[3, 3],
    legend=False,
    figsize=[12, 8],
    colormap="Accent",
    title="Show all metrics in Accent colormap",
)

# Customize plots with kind (note that we are only plotting the "count" metric here because we are showing a pie chart)
metric_frame.by_group[["count"]].plot(
    kind="pie",
    subplots=True,
    layout=[1, 1],
    legend=False,
    figsize=[12, 8],
    title="Show count metric in pie chart",
)

# Saving plots
fig = metric_frame.by_group[["count"]].plot(
    kind="pie",
    subplots=True,
    layout=[1, 1],
    legend=False,
    figsize=[12, 8],
    title="Show count metric in pie chart",
)

# Don't save file during doc build
if "__file__" in locals():
    fig[0][0].figure.savefig("filename.png")

• 
• 
• 
• 
• 

/home/circleci/tmp-fairlearn/examples/plot_quickstart.py:27: SettingWithCopyWarning:
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  X.drop(columns=["readmitted", "readmit_binary"], inplace=True)
/home/circleci/.pyenv/versions/3.11.5/lib/python3.11/site-packages/sklearn/metrics/_classification.py:1509: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, f"{metric.capitalize()} is", len(result))
/home/circleci/.pyenv/versions/3.11.5/lib/python3.11/site-packages/sklearn/metrics/_classification.py:1509: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, f"{metric.capitalize()} is", len(result))
/home/circleci/.pyenv/versions/3.11.5/lib/python3.11/site-packages/sklearn/metrics/_classification.py:1509: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, f"{metric.capitalize()} is", len(result))
/home/circleci/.pyenv/versions/3.11.5/lib/python3.11/site-packages/sklearn/metrics/_classification.py:1509: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, f"{metric.capitalize()} is", len(result))