General compression techniques

[MasterInQuestion]

    On general data compression, there are 3 possible techniques:
    |1| Collapse identical: [
    Merge identical values into position description shorter than raw original.
    Common techniques: RLE (Run-Length Encoding), Delta compression, Lempel-Ziv (LZ algorithms), Huffman (frequency optimized prefix coding).
    Note Huffman conceptually is rearranging the symbols' representation based on frequency.
    But fundamentally alike. ("symbols" are mere conceptual) ]
    .
    |2| Value interpolation: [
    Interpolating (transform) the intermediate values based on certain pattern.
    The most classical example is "1, 2, 3, 4, 5, ..." alike sequence.
    Suitable for gradient alike. ]
    .
    |3| Keep as-is: [
    For data cannot be compressed. (random data alike)
    For true stochastic data: the ideal compression strategy is no compression at all.
    (for no redundancy exploitable) ]

    The key distinction between "Collapse identical" and "Value interpolation" is:
    "Copy" vs. "Calculate"