ForgeStat

General-purpose statistics engine. Pure Python computation with numpy/scipy.

Install

pip install forgestat

Quick Start

from forgestat.parametric.ttest import one_sample_t, two_sample_t
from forgestat.nonparametric.rank_tests import mann_whitney_u
from forgestat.regression.linear import ols_regression
from forgestat.posthoc.comparisons import tukey_hsd

result = two_sample_t([1, 2, 3, 4], [2, 3, 4, 5])
mw = mann_whitney_u([1, 2, 3], [4, 5, 6])

Modules

Module	Contents
core	Result types, assumption checks, effect sizes, distribution fitting
parametric	t-tests, ANOVA, correlation, chi-square, proportions, equivalence, repeated measures, split-plot, variance tests
nonparametric	Mann-Whitney, Kruskal-Wallis, Wilcoxon, Friedman, runs test
posthoc	Tukey HSD, Dunnett, Games-Howell, Dunn, Bonferroni, Scheffe
regression	OLS, logistic, Poisson, nonlinear, stepwise, GLM, robust, best subsets, ordinal, orthogonal
bayesian	Bayesian t-test, ANOVA, proportion, correlation, regression
exploratory	Descriptive stats, PCA, MANOVA, meta-analysis, multi-vari
power	Sample size for t, ANOVA, chi-square, proportion, regression, variance, DOE, tolerance
quality	Process capability, acceptance sampling, ANOM, variance components
reliability	Kaplan-Meier, Weibull, Cox PH, exponential, lognormal
msa	Gage R&R, ICC, Bland-Altman, linearity/bias, Krippendorff alpha
timeseries	ADF, KPSS, ACF/PACF, ARIMA, SARIMA, decomposition, changepoint, Granger causality

Dependencies

• numpy, scipy (required)
• statsmodels (optional, for select advanced methods)

License

MIT