(redirected from Downsample)
Also found in: Dictionary, Thesaurus, Medical, Encyclopedia.


the selection of part of a total population of consumers or products whose behaviour or performance can be analysed, in order to make inferences about the behaviour or performance of the total population, without the difficulty and expense of undertaking a complete census of the whole population.

Samples may be chosen randomly, with every consumer or product in the population having an equal chance of being included. Random samples are most commonly used by firms in QUALITY CONTROL where they are used as a basis for selecting products, components or materials for quality testing.

Alternatively, samples may be chosen by dividing up the total population into a number of distinct sub-groups or strata, then selecting a proportionate number of consumers or products from each sub-group since this is quicker and cheaper than random sampling. In MARKETING RESEARCH and opinion polling, quota sampling is usually employed where interviewers select the particular consumers to be interviewed, choosing the numbers of these consumers in proportion to their occurrence in the total population.

Samples may be:

  1. cross-sectional, where sample observations are collected at a particular point in time, for example data on company sales and the incomes of consumers in the current year, embracing a wide range of different income groups, as a basis for investigating the relationship between sales and income;
  2. longitudinal, where sample observations are collected over a number of time periods, for example data on changes in company sales over a number of years and changes in consumer incomes over the same time periods, as a basis for investigating the relationship between sales and income. See STATISTICAL INFERENCES, QUESTIONNAIRE.
References in periodicals archive ?
We collect around 80 images of baseball players and randomly upsample or downsample all the collected images to create a dataset.
Previously this algorithim designed with different subsampled (oversampled or downsample and upsample) factors like 4,8,12,24 and so on upto 33.
Once we find this optimum tradeoff, we can adjust the camera setting or downsample the input data.
After implementing the optimizations described in section 4, we concluded that the best practice to obtain the lower error in the shorter time is to downsample the picture and to reduce the points, only keeping the middle of each entity.
IIP eliminates the need to manually alter every image published since it can essentially downsample "on the fly" while allowing users to zoom in on the images they want to see in more detail.