Module testing 

Hypothesis testing.

Parametric and non-parametric statistical tests: t-tests, ANOVA, chi-squared tests, and normality tests.

Examples

use u_analytics::testing::{one_sample_t_test, TestResult};

let data = [5.1, 4.9, 5.2, 5.0, 4.8, 5.3, 5.1, 4.9];
let result = one_sample_t_test(&data, 5.0).unwrap();
assert!(result.p_value > 0.05); // cannot reject H₀: μ = 5.0

Structs

AdfResult

Result of the Augmented Dickey-Fuller (ADF) unit root test.

AndersonDarlingResult

Result of the Anderson-Darling normality test.

AnovaResult

Result of one-way ANOVA.

KpssResult

Result of the KPSS stationarity test.

MannKendallResult

Result of the Mann-Kendall trend test.

ShapiroWilkResult

Result of the Shapiro-Wilk normality test.

TestResult

Result of a hypothesis test.

Enums

AdfLags

How adf_test chooses the number of lagged differences.

AdfModel

Model specification for the ADF test.

KpssLags

The truncation lag of the Newey-West long-run variance in kpss_test.

KpssModel

What the KPSS null hypothesis says the series is stationary around.

KpssPValue

Where a KpssResult::p_value stands relative to the table it is read from.

Constants

ADF_MIN_OBSERVATIONS

The fewest values adf_test takes.

KPSS_LEVELS

Significance levels of KpssResult::critical_values, in order.

KPSS_MIN_OBSERVATIONS

The fewest values kpss_test takes.

Functions

adf_lags_domain

The numbers of lagged differences adf_test takes on data under model – for a fixed count (lags) and for the bound of the AIC search (max_lags) alike: 0..=p, where with p lags the regression keeps n − 1 − p rows (one value is lost to differencing, p to the lags) and has the lagged level, the model’s deterministic terms and p lag columns; it needs at least 5 rows and a residual degree of freedom.

adf_test

Augmented Dickey-Fuller (ADF) unit root test for stationarity.

anderson_darling_test

Anderson-Darling normality test: H₀: data is normally distributed.

bartlett_test

Bartlett test for equality of variances: H₀: all groups have equal variance.

benjamini_hochberg

Benjamini-Hochberg FDR correction.

bonferroni_correction

Bonferroni correction: adjusts p-values for multiple comparisons.

chi_squared_goodness_of_fit

Chi-squared goodness-of-fit test: H₀: observed matches expected distribution.

chi_squared_independence

Chi-squared test of independence on a contingency table.

fisher_exact_test

Fisher exact test for a 2×2 contingency table.

jarque_bera_test

Jarque-Bera normality test: H₀: data is normally distributed.

kpss_lags_domain

The values kpss_test takes for a fixed truncation lag on data: 0..=n − 1.

kpss_test

KPSS test of stationarity: H₀ the series is stationary around a level (or a trend), H₁ it has a unit root.

kruskal_wallis_test

Kruskal-Wallis test: H₀: all groups have the same distribution.

levene_test

Levene test for equality of variances: H₀: all groups have equal variance.

mann_kendall_test

Mann-Kendall non-parametric trend test with Sen’s slope estimator.

mann_whitney_u_test

Mann-Whitney U test: H₀: the two populations have the same distribution.

one_sample_t_test

One-sample t-test: H₀: μ = μ₀.

one_way_anova

One-way ANOVA: H₀: all group means are equal.

paired_t_test

Paired t-test: H₀: mean difference = 0.

shapiro_wilk_test

Shapiro-Wilk normality test: H₀: data is normally distributed.

two_sample_t_test

Two-sample Welch t-test: H₀: μ₁ = μ₂ (unequal variances).

wilcoxon_signed_rank_test

Wilcoxon signed-rank test: H₀: median of differences = 0.