The approximation of a numeric data item by its nearest equivalent, to a given number of significant figures or decimal places. For example, 1.36 rounded to one decimal place is 1.4.