Autocorrelation

Autocorrelation — does your time series echo its own past? A tiny, offline, beautifully designed tool for autocorrelation analysis. Plot the ACF and PACF correlograms with significance bounds, run the Ljung-Box white-noise test, and instantly see whether a series carries momentum, cycles or seasonality — no internet, no accounts, no clutter. Whether you are a student learning time-series analysis, a data analyst sanity-checking residuals, or just curious why some numbers seem to repeat, this app turns a column of values into a clear, readable fingerprint. WHAT YOU CAN DO • Edit your own time series — type values manually, nudge them with + / − buttons, or drag a slider. Your data is never lost. • Set the maximum lag and watch every result update live. • Read the ACF (autocorrelation function) correlogram, with significant bars highlighted against the ±1.96/√n significance band. • Read the PACF (partial autocorrelation function), computed by the Durbin-Levinson recursion, to see the direct link at each lag. • Get the Ljung-Box Q statistic and its p-value, with a plain-language verdict: white noise, or real structure? • View the raw series as a clean line chart. FIVE FOCUSED TABS 1. Concept — what autocorrelation actually is, ACF vs PACF, the significance band and the Ljung-Box test, explained simply with the formulas rendered in crisp KaTeX (fully offline). 2. Data — the editable time-series with a max-lag control; see the lag-1 ACF, the strongest significant lag, the Ljung-Box Q and p-value, and a verdict. 3. ACF — the ACF and PACF correlograms side by side, plus the raw-series line. 4. Simulate — turn the knobs for length, AR(1) momentum, seasonal period and noise, and watch the correlogram fingerprint of each structure appear. 5. Examples — ready-made series (seasonal, strong momentum, white noise, alternating) that load with a single tap. WHY YOU WILL LIKE IT • 100% offline — your data stays on your device. No network permission needed. • No external math or chart libraries. The sample ACF, the PACF by Durbin-Levinson, the significance bounds and the Ljung-Box chi-square p-value are all computed from scratch; the correlograms are drawn pixel by pixel. The result is a tiny app (~3 MB). • A calm, modern "Rose Quartz" design built with Material 3. • Learn by doing — the Concept tab and the Simulate tab make the theory click. GREAT FOR • Students of statistics, econometrics and signal processing. • Analysts checking whether residuals are white noise. • Anyone exploring momentum, cycles and seasonality in data. WHAT IS AUTOCORRELATION? Autocorrelation measures how much a series resembles a shifted copy of itself. A high autocorrelation at lag k means a value tends to look like the value k steps earlier — the signature of momentum, a cycle, or a season. The ACF shows this at every lag; the PACF isolates the direct effect of each lag; the significance band tells you which spikes are real; and the Ljung-Box test bundles it all into a single yes/no on whether the series is just noise. No sign-up, no clutter — just a focused tool that does one thing well. Supported by a single, unobtrusive banner ad.