# What’s the difference between ‘decimated‘ and ‘undecimated’ data?

How is ‘decimated’ data different from ‘undecimated’ data?

## 1 Answer

"Decimated" is another word for "downsampling," or in other words resampling the original ("undecimated") data with a lower sampling rate. So the decimated data will have a lower sampling rate than the undecimated data.

Edit As pointed out in the comments, it is a very good idea to low pass filter the data before downsampling, to avoid aliasing effects. There seem to be different conventions as to whether low pass filtering is included in the definition of "decimation".

• There's a subtle difference between the two isn't there? I believe that decimation is low-pass filtering, e.g., CIC filter, followed by downsampling. For example, the Matlab function downsample(x,n) keeps every nth sample from x, while the function decimate(x,r) low-pass filters x first and then keeps every rth sample Jul 3, 2020 at 17:04
• The way I learned it, decimating refers to downsampling, but it is a good idea to low pass filter before downsampling. (Eg, on wikipedia (en.wikipedia.org/wiki/Downsampling_(signal_processing)): "Decimate the filtered signal by M; that is, keep only every Mth sample.)" However, this is just semantics, one should certainly low pass filter first. Jul 3, 2020 at 17:10
• In Oppenheim's Discrete-Time Signal Processing, a downsampler that does not pre-filter the input is called a sampling rate compressor or compressor, while a downsampler that does low-pass filter the input is called a decimator.
– Puk
Jul 3, 2020 at 17:55