stats_correlogram | Autocorrelation (ACF) and partial autocorrelation (PACF) of one series up to nlags, each with its confidence band, plus the autocovariances and the Ljung-Box and Box-Pierce Q statistics with p-values at every lag. | |
stats_cross_correlogram | Cross-correlation of two equally long series at lags −nlags to +nlags: the value at lag k is corr(x[t+k], y[t]), so a peak at k > 0 means y leads x by k periods and at k < 0 that x leads y. | |
stats_seasonal_decompose | Splits a series into trend, seasonal and residual components by STL (LOESS, default), classical moving averages (additive or multiplicative) or MSTL (several seasonal periods at once). | |
stats_trend_cycle_filter | Separates a series into trend and cycle with the Hodrick-Prescott filter, the Baxter-King or Christiano-Fitzgerald band-pass filters (cycles between low and high periods), or a polynomial detrend. | |
stats_periodogram | Spectral density of a series by the periodogram or Welch’s averaged periodogram. | |
stats_stationarity_test | Unit-root and stationarity tests on one series: augmented Dickey-Fuller (null: unit root), KPSS (null: stationary), both together with a joint verdict (default), Zivot-Andrews (unit root against stationarity with one structural break, whose date it reports) and the range unit-root test. | core |
stats_engle_granger | Engle-Granger two-step cointegration test of y on one to five x columns: an ADF test on the residual of the cointegrating regression with MacKinnon p-values (null: no cointegration). | |
stats_bds_test | BDS test of the null that a series is independent and identically distributed, at embedding dimensions 2 to max_dim. | |
stats_breakvar_test | Test of a change in variance between the first and last parts of a series (usually model residuals): the ratio of the sums of squares of the last and first subsets. | |
