Which forecasting method uses a weighted average of past time series values that emphasizes the most recent observation?

The forecasting method that uses a weighted average of past time series values, specifically placing greater emphasis on the most recent observations, is exponential smoothing. This technique is designed to give more weight to recent data points while still considering older observations, but the impact of those older data points diminishes over time.

In exponential smoothing, the most recent observation influences the forecast more significantly than those further back in the time series, allowing for a more responsive approach to data that can fluctuate over time. This is particularly beneficial in situations where trends or patterns change rapidly, as it helps the forecast to react more quickly to those changes.

The other methods differ in their approaches. For instance, a simple moving average treats all past observations equally, providing an average over a set period without giving more weight to the most recent data. A weighted moving average does allow for the assignment of different weights to observations but may not necessarily emphasize the most recent observation as explicitly as exponential smoothing does. Linear regression, on the other hand, is a predictive analytical method that establishes a linear relationship between variables but does not focus on time series forecasting in the same manner.