Generate imbalance multisite data

This example shows how to generate an unbalanced multisite dataset using the balance_per_site parameter of the make_multisite_classification function.

Imports

import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns

from uniharmony import verbosity
from uniharmony.datasets import make_multisite_classification


sns.set_theme(style="whitegrid")
verbosity("warning")

Data generation

Let’s start with the function as default, this will create a 2 site balanced problem.

X, y, sites = make_multisite_classification()
df = pd.DataFrame({"Class": y, "Site": sites})

general_balance = len(y[y == 1]) / len(y)
y_site_0 = y[sites == 0]
y_site_1 = y[sites == 1]
site_0_balance = len(y_site_0[y_site_0 == 1]) / len(y_site_0)
site_1_balance = len(y_site_1[y_site_1 == 1]) / len(y_site_1)

print(
    "The class distribution is balanced across sites and in general \n"
    f"General balance: {general_balance:.2f} \n"
    f"site 0 balance: {site_0_balance:.2f} \n"
    f"site 1 balance: {site_1_balance:.2f}"
)

plt.figure(figsize=[10, 6])
plt.title("Class and site distribution")
sns.countplot(df, x="Class", hue="Site")
plt.grid(axis="y", color="black", alpha=0.5, linestyle="--")

The class distribution is balanced across sites and in general
General balance: 0.50
site 0 balance: 0.50
site 1 balance: 0.50

Let’s now create a site imbalance problem. That means that, while the total number of examples per class is imbalance, the classes are not equally distributed by site.

X, y, sites = make_multisite_classification(balance_per_site=[0.3, 0.7])
df = pd.DataFrame({"Class": y, "Site": sites})
general_balance = len(y[y == 1]) / len(y)
y_site_0 = y[sites == 0]
y_site_1 = y[sites == 1]
site_0_balance = len(y_site_0[y_site_0 == 1]) / len(y_site_0)
site_1_balance = len(y_site_1[y_site_1 == 1]) / len(y_site_1)

print(
    "The class distribution is imbalanced across sites but balanced in general \n"
    f"General balance: {general_balance:.2f} \n"
    f"site 0 balance: {site_0_balance:.2f} \n"
    f"site 1 balance: {site_1_balance:.2f}"
)

plt.figure(figsize=[10, 6])
plt.title("Class and site distribution")
sns.countplot(df, x="Class", hue="Site")
plt.grid(axis="y", color="black", alpha=0.5, linestyle="--")

The class distribution is imbalanced across sites but balanced in general
General balance: 0.50
site 0 balance: 0.30
site 1 balance: 0.70

X, y, sites = make_multisite_classification(balance_per_site=[0.3, 0.3])
df = pd.DataFrame({"Class": y, "Site": sites})
general_balance = len(y[y == 1]) / len(y)
y_site_0 = y[sites == 0]
y_site_1 = y[sites == 1]
site_0_balance = len(y_site_0[y_site_0 == 1]) / len(y_site_0)
site_1_balance = len(y_site_1[y_site_1 == 1]) / len(y_site_1)

print(
    "The class are imbalanced in general, but have the same imbalance across sites\n"
    f"General balance: {general_balance:.2f} \n"
    f"site 0 balance: {site_0_balance:.2f} \n"
    f"site 1 balance: {site_1_balance:.2f}"
)

plt.figure(figsize=[10, 6])
plt.title("Class and site distribution")
sns.countplot(df, x="Class", hue="Site")
plt.grid(axis="y", color="black", alpha=0.5, linestyle="--")

2026-05-18 13:04:34 [warning  ] Not enough samples of class 0 in global dataset. Requested 350, available 330. Consider adjusting balance_per_site or generating more samples.
2026-05-18 13:04:34 [warning  ] Not enough samples of class 0 in global dataset. Requested 350, available 330. Consider adjusting balance_per_site or generating more samples.
The class are imbalanced in general, but have the same imbalance across sites
General balance: 0.30
site 0 balance: 0.30
site 1 balance: 0.30