An autoregressive model predicts the next observation from earlier observations. In AR($p$), the next value depends on the last $p$ values and a fitted set of coefficients.

That makes it useful for quick short-horizon forecasting, pattern checks, and comparing how strongly the recent past influences the next step.

Add your series in order, choose a lag order, and apply the model. The chart separates the observed history from the projected tail so you can inspect whether the fitted trend is plausible.

If the fit fails, the data may not have enough variation for the selected lag order. Reduce the lag or add more history points.