Prediction method
In electronic data processing , prediction methods provide estimated values for values of a data set on the basis of other data in the set using redundancy.
use
- In data compression , they are used so that after the forecast for a value only the usually smaller difference (residual value) to the forecast value has to be coded ( delta coding ), which can usually be represented more compactly at least after entropy coding .
- In signal processing, for example, prediction methods usually serve as the basis for lossless compression methods for audio signals, for example (see audio data compression ).
The better the prediction, the more chaotic and closer to white noise the residual signal will be. The more predictable the output signal, the lower the level of the residual signal, which can be represented more compactly after entropy coding. The residual values can also be quantized in order to achieve a data reduction .
- A common method is, for example, linear prediction , which is used in communications technology to compress a signal data stream by means of linear predictive coding (LPC) based on it.