On Thu, 17 Feb 2022 at 21:20, Thomas Munro <thomas.munro@xxxxxxxxx> wrote:
On Fri, Feb 18, 2022 at 9:11 AM Shaozhong SHI <shishaozhong@xxxxxxxxx> wrote:
> How to calculate frequency of positive and negative numbers and define and calculate frequency of alteration of polarity?
>
> Surely, we can use frequency of alteration of polarity and level of change (e.g., size of positive and negative numbers) to measure degree and frequency of alteration.
>
> Any ideas in doing so in postgres tables' columns full of positive and negative numbers?
Window functions might be useful to detect polarity changes:
postgres=# create table time_series (time int, value int);
CREATE TABLE
postgres=# insert into time_series values (1, -5), (2, -5), (3, 10), (4, -3);
INSERT 0 4
postgres=# select time,
value,
sign(lag(value) over (order by time)) != sign(value)
as flipped
from time_series;
time | value | flipped
------+-------+---------
1 | -5 |
2 | -5 | f
3 | 10 | t
4 | -3 | t
(4 rows)
Given 2 or more such columns, is there any measure that can be calculated to tell which one alternates more than others?
Regards,
David
