This article demystified data compression. Essentially, data compression involves "taking a representation of information and replacing it with a different representation that takes up less space and from which the original can be recovered."
With lossless compression, the recovered info is guaranteed to be exactly the same as the original. This type of compression is recommended for computer programs, because every bit is important to the functioning of the program.
Lossy compression, on the other hand, preserves meaning of data rather than the data. The recovered info is never identical to the original. This type of compression is useful for info that is meant to be interpreted by a "meaning processor" such as a human being or an image recognition system. Uses of lossy compression could be to take the hiss out of a voice recording, or to store photographs.
Although this article did an excellent job of explaining the different data compressions, I believe the examples used and some of the technical information only palatable to someone interested in sound and video engineering.