Sometimes we have so much data it cannot be processed in its entirety, in which case some form of data reduction is necessary. This may be about reducing the number of data items and/or reducing the size of each item. The number of data items may be reduced by filtering so that only a subset of the data items are used, or in some way grouping items, for example working with area averages for geographic data. Similarly we might choose to only use a subset of features in each data item, combine features, or process complex features (such as images).
Data reduction methods. Fig 11.1 from Statistics for HCI (2e).