Time Series Calculators

Sample autocorrelation (ACF) and partial autocorrelation (PACF, Durbin-Levinson) of a series up to lag k, with 95% bounds.

Find level shifts in a series with two-sided CUSUM alarms and a mean-shift binary segmentation, positions 1-based.

Simple, Holt trend or Holt-Winters seasonal exponential smoothing with a forecast h steps ahead and in-sample fit error.

Score a forecast against actuals: MAE, RMSE, MAPE, sMAPE, bias and MASE against a naive benchmark.

Period-over-period percent change, log returns, CAGR and rolling growth for a list of values in time order.

Simple, exponential or weighted moving average of a list of numbers, trailing or centered, with the alignment stated.

Flag outliers in a list of numbers by z-score, Tukey IQR fences and the MAD modified z-score (3.5 rule).

Classical additive decomposition of a series into trend, seasonal and remainder parts for a given period.