A method of smoothing a time series. Let x1, x2,…, xn denote the observed values, and let x̂1, x̂2,…, x̂n be a corresponding set of smoothed values computed using the formulae
where α (0<α<1) is a constant. An equivalent formula is
The value of α is usually chosen to minimize With this value of α, x̂t is called the exponentially weighted moving average (EWMA).With this value of α, the formula may be adapted for forecasting a time series having neither trend (see moving average) nor seasonality by writing
where x̂t + 1 is the predicted value for time (t + 1) and it is assumed that a long series of values exists.
A forecast based on the weighted average of the past values of the variable, in which the weights assigned to the past values decline exponentially with the lag length. This procedure is also used for data smoothing and is a version of the weighted moving average technique.
- Ramsey pricing
- Ramsey regression equation specification error test
- Ramsey rule
- Ramsey sentence
- Ramsey test
- Ramsey Theory
- Ramsey–Bordé interferometer
- ramus communicans
- Ramus, Petrus (1515–82)
- Ramón y Cajal, Santiago
- Randall, Sir John Turton
- Randall–Sundrum scenario
- Rand, Ayn (1905–82)
- RAND corporation
- Randian
- Randolph, Asa Philip (1889–1979)
- random
- random access
- random-access file
- random-access memory
- random-access stored-program machine
- random algorithms
- random alloy
- random cascade model
- random coefficient model