Time-Weighted Averages Blog Questions & Feedback

[davidkohn88]

This is a discussion for folks who have questions / comments / feedback on our recent blogpost on time-weighted averages: https://blog.timescale.com/blog/what-time-weighted-averages-are-and-why-you-should-care/

[abdelkaderralem]

Hi Everyone,
I have read "What Time-Weighted Averages Are and Why You Should Care" post.
I m doing simulation about TCP protocol.
I have a two column data time and the value for the instant throughput of tcp.
could i use the function to calculate the time-weighted average for throughput, and that for every instance when the value change.
as a result i will get the time-weighted average of TCP throughput.
Thank You.

[Timsgmlr]

I'm a little confused about the behavior of time weighted averages using LOCF. Let's say I have following data points:

Time	Value
2020-01-01 00:00:00	1
2020-01-01 23:00:00	1000

When calculating the weighted average for a 24 hour interval my calculation would be: 1 * 23/24 + 1000 * 1/24 = 42.625.
When doing the same thing with timescaledb using following query:

SELECT time_bucket('1 days', time) AS timebucket,
       average(time_weight(
               'locf',
               time,
               value
           ))                      
FROM series_points_float
WHERE (time BETWEEN TIMESTAMP '2020-01-01 00:00:00+00:00' AND TIMESTAMP'2020-01-02 00:00:00+00:00')
GROUP BY timebucket

The result is just 1. Can someone please point out why the second data point is not taken into consideration?