Percentiles may appear in performance testing reports. This topic describes percentiles and how to calculate percentiles.
What are percentiles?
Percentiles provide a clearer picture of the overall distribution of user data and minimize the influence of anomalies. Unlike the mean, which can be skewed by outliers, high percentiles disregard a small number of extreme values to provide a more stable measure.
High percentile data may not be suitable for all business contexts. For example, in the financial payment industry, all transactions may require a 100% success rate.
To obtain a percentile, sort data in ascending order and then identify the value at the position that corresponds to the desired percentage.
How are percentiles calculated?
For example, to obtain the 95th percentile from a collected dataset of 100 data points, arrange the data points in ascending order and identify the 95th data point in the ordered list.
Similarly, the 50th percentile value is the data of the 50th data point.