The range quantifies the maximum spread of a dataset by measuring the absolute gap between its extreme endpoints.
It is calculated by taking the maximum observed value and subtracting the minimum observed value.
For example, if a temperature sensor records a peak of thirty-two degrees and a trough of eight degrees, the range is twenty-four degrees.
Range provides a rapid sanity check during early exploratory analysis to verify whether numbers fall within realistic physical limits.
In industrial quality control, sample ranges on small batches are plotted over time to detect sudden shifts in machine tolerances.
However, the range uses only two observations out of the entire dataset, discarding all intermediate information.
Two datasets with identical ranges can have completely different internal structures and spread characteristics.
One dataset may cluster ninety-nine percent of its values around the centre, while another spreads values uniformly across the span.
Because it depends directly on the extreme sample minimum and maximum, the range tends to increase as sample size grows.
Larger samples provide more opportunities to capture rare extreme events, naturally expanding the observed sample range.
A single corrupted measurement, such as an extra zero typed during manual data entry, distorts the range completely.
This extreme vulnerability makes the range an unreliable metric for formal statistical hypothesis testing and variance comparison.
Despite its instability, range remains useful for setting bounding limits in data visualisations and normalising features into fixed intervals.
In min-max feature scaling, the range forms the denominator that scales raw variables into the interval between zero and one.
Analysts should always pair the range with interior spread metrics like the interquartile range to understand dispersion properly.