No abstract

The suitability of a new method to determine the minimum number of data required to stabilize a calculation is studied. Five different calculations have been computed, changing the order of data randomly. The statistical reliability of numerical values has also been calculated depending on the successive increase in the number of data included. In our study, when few data are considered, a great variation in the estimations is observed depending on the individual data included. The calculation is stabilized for a sample size between 5 and 15 data points, and the sample can therefore be considered representative enough. Results are in agreement with those presented by other authors.

Abstract-The suitability of a new method to determine the minimum number of data required to stabilize a calculation is studied. Five different calculations have been computed, changing the order of data randomly. The statistical reliability of numerical values has also been calculated depending on the successive increase in the number of data included. In our study, when few data are considered, a great variation in the estimations is observed depending on the individual data included. The calculation is stabilized for a sample size between 5 and 15 data points, and the sample can therefore be considered representative enough. Results are in agreement with those presented by other authors.