Percentile based on already aggregated data

Elastic Stack Elasticsearch
1

Hi all,

There is a requirement from my project that we need to compute percentiles based on aggregated data. For instance we have the following data records:
throughput count
100 5
200 1
800 4

So ideally, we would like to compute percentile by treating the data as the following:
100, 100, 100, 100, 100, 200, 800, 800, 800, 800

I wonder if ES can help in this case?

I searched in the forum and found there was a post from 2 years ago asking the same question, unfortunately there was no reply and seems like the post was closed.

Thanks in advance!
Bin

2

One way to solve this would be to use a script in the percentiles aggregation. Your script could output an array of length count in which every element is throughput and this should have the affect you want.

3

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.