Linear Regression Type II for Python

This python package (pylr2) provides linear regression type 2 (regress2), which are recommended when there is variability in both variables regressed. It uses linear regression type 1 from the statsmodels package.

Linear Regression Type 1 models supported are:

• OLS: ordinary least square <default>
• WLS: weighted least square
• RLM: robust linear model

Linear Regression Type 2 models supported are:

• major axis
• reduced major axis ( also known as geometric mean) <default>
• arithmetic mean

Example of usage:

from pylr2 import regress2


# Generate a dataset
x = np.linspace(0, 10, 100)
e = np.random.normal(size=len(x))
y = x + e
# Compute regression type 2
results = regress2(x, y, _method_type_2="reduced major axis")

The package is an adaptation of the type 2 models from MBARI (https://www.mbari.org/products/research-software/matlab-scripts-linear-regressions/).