Statistics & Machine Learning¶
Regression¶
The canonical wrapper supports OLS, quantile, logistic, polynomial, factor and regularized regression variants.
Walk-forward machine learning¶
wf = asr.ml.fit(
"ridge",
features,
target,
train_size=120,
test_size=30,
gap=1,
)
print(wf.summary)
The walk-forward API keeps training and testing chronological. A gap can be used when the target horizon or feature construction requires an embargo between training and evaluation windows.
Statistical research utilities¶
The broader statistics namespace includes rolling regression, stationarity diagnostics, block bootstrap, permutation tests, Benjamini-Hochberg multiple-testing control, cointegration, Granger causality and time-series model helpers.