Television pictures contain a lot of information and often have to be reduced before being recorded or transmitted. This is partly done by a technique known as compression. Compression techniques take away some of the information which is either redundant or has little effect on how we percieve the compressed image. Compression techniques are more aggressive if the final viewing platform is say a mobile phone and far less aggressive for transmission as a broadcast quality high definition television signal .
Compression ratio is a measure of the amount of compression taking place and refers to the relationship between the number of bits entering a 'compression engine' and the number of bits coming out the other side.
So if a fully sampled uncompressed 270 Mbps signal goes into a compressor and 50 Mbps comes out of the other side, then the compression ratio is 270/50 or about 5.5:1
Other techniques for reducing the amount of data are bit depth reduction and chroma subsampling. These simply remove some data and do not cause any compression. The combined effect of data reduction and compression is know as a codec.
Now - Lots of codecs use all 3 techniques - The DV codec for example goes from 10 bits sampling to 8 (bit depth reduction) then 4:2:2 to 4:2:0 (chroma sub-sampling) and then compresses the result to create an output bit rate of 25Mbps. - so the ratio of input bits to output bits is 270/25 or close to 11 to 1. (Partly 'data reduction' and partly compression).
Most people will just say DV is about 10:1 compressed - and frankly that is good enough for most purposes.
So when we talk about compression or compression ratio, it doesn't just refer to the compression part of the codec but rather the overall combined effect of data reduction and compression. Simply the ratio of what goes in to what comes out.