def get_adult(shuffle=True, seed=None):
np.random.seed(seed)
adult = fetch_adult()
X = adult["data"]
y = adult["target"]
X[:, [9,-1]] = X[:, [-1,9]] # puts sex into the last column
# Erase nan values
y = y[~np.isnan(X).any(axis=1)]
X = X[~np.isnan(X).any(axis=1)]
# shuffle data
if shuffle:
permute = np.random.permutation(len(y))
y = y[permute]
X = X[permute]
# transform to binary and set proper type
X = X.astype('float')
for i, lab in enumerate(y):
if y[i] == "<=50K":
y[i] = 0.
else:
y[i] = 1.
y = y.astype('float')
return X, y
def get_adult():
adult = fetch_adult()
X = adult["data"]
y = adult["target"]
X[:, [9, -1]] = X[:, [-1, 9]] # puts sex into the last column
# Erase nan values
y = y[~np.isnan(X).any(axis=1)]
X = X[~np.isnan(X).any(axis=1)]
# transform to binary and set proper type
X = X.astype('float')
for i, lab in enumerate(y):
if y[i] == "<=50K":
y[i] = 0.
else:
y[i] = 1.
y = y.astype('float')
n, _ = X.shape
return X, y
# Copyright (c) Microsoft Corporation and Fairlearn contributors.
# Licensed under the MIT License.
"""Produce plot of selection rates for the quickstart guide."""
from bokeh.plotting import figure, show
from fairlearn.metrics import MetricFrame, selection_rate
from fairlearn.datasets import fetch_adult
data = fetch_adult(as_frame=True)
X = data.data
y_true = (data.target == '>50K') * 1
sex = X['sex']
selection_rates = MetricFrame(selection_rate,
y_true,
y_true,
sensitive_features=sex)
xs = list(selection_rates.by_group.index)
ys = [selection_rates.by_group[s] for s in xs]
p = figure(x_range=xs,
plot_height=480,
plot_width=640,
title="Fraction earning over $50,0000",
toolbar_location=None,
tools="")
p.vbar(x=xs, top=ys, width=0.9)
p.y_range.start = 0
p.xgrid.grid_line_color = None