ChiSquared¶

Streaming Chi-squared statistic.

Maintains a contingency table between two variables x and y and computes the Chi-squared statistic incrementally. This can be used to measure the dependency between two categorical variables in a streaming setting.

Attributes¶

• degrees_of_freedom

  Return the degrees of freedom of the contingency table.

• name

• p_value

  Return the p-value associated with the Chi-squared statistic.

Examples¶

from river import stats

chi2 = stats.ChiSquared()

data = [
    ("A", 0),
    ("A", 0),
    ("B", 1),
    ("B", 1),
]

for x, y in data:
    chi2.update(x, y)

round(chi2.get(), 3)

4.0

A rolling version can be obtained by wrapping with utils.Rolling:

from river import utils

data = [
    ("A", 0),
    ("A", 0),
    ("B", 1),
    ("B", 1),
    ("C", 0),
]

chi2 = utils.Rolling(stats.ChiSquared, window_size=4)

for x, y in data:
    chi2.update(x, y)

round(chi2.get(), 3)

4.0

Methods¶