|standard or uniform distribution
||distribution explicitly investigated
||control character for new symbols
|Distribution located at encoder and decoder as well.
||Distribution has to be transported within the header.
||No additional data necessary.
|Compression efficiency at the beginning|
|High; provided that the assumed distribution matches the real data.
||High; provided that the initial distribution does not differ from the entire set of data.
||Low; because all symbols has to be introduced by the control character.
|Tendency of the compression efficiency in the course of the process|
|Constant; provided that the assumed distribution matches the real data.
||High; provided that no strong deviations from the chosen distribution appear locally.
||High; with progressive coding the compression efficiency improves continuously in contrast to the other algorithms.
|Compression efficiency for larger amounts of data|
|Depending on the accordance with the standard distribution.
||Depending on the accordance with the chosen distribution.
||High; because the current code tree is adapted to the local distribution.
|Fast; because no code tree has to be generated.
||Fast; because the code tree has to be generated only once (or periodically).
||Slow; because the code tree has to be adapted for each symbol.