<_rudi> when collecting data into a frequency table of n-bits, i wonder if its better to expand the table to n+1 bits and so on, so that the frequencies have higher differences?
<_rudi> that is for compressing data
<_rudi> if one for example wants to do delta-coding
<_rudi> reading two bits for example makes the data collection more dense, while increasing it to more bits makes the data more dispersed. if one does delta coding on 2 bits i guess it wont compress much, but doing it for more bits, if the rawdata is large will help?
<_rudi> also, another question is when should one do deltacoding? for example here's different frequency tables for the same dataset: http://pastebin.com/raw/WQvaKGb0
<_rudi> dcoding on 2 bits doesnt seem right, not 3 bits either. but on 4,5 or 6? 5-bits have one empty slot. 6-bits have several empty slots. and when should one stop. or is it some other feature that one should look at. ive read that dcoding works well for low-frequency data: that is where the difference between each element is low. so increasing the bit-size for dense data doesnt seem wrong.
<_rudi> "Delta encoding can be used for data compression when the values in the original data are smooth, that is, there is typically only a small change between adjacent values."