Re: Looking for "gaps" (from a time perspective) in data

On Oct 10, 4:27 pm, jeremy <jeremy0..._at_gmail.com> wrote:
> On Oct 10, 1:04 pm, ca111..._at_gmail.com wrote:
>
>
>
>
>
> > Hi Jeremy,
>
> > Instead of looking at individual records it may be easier to analyze
> > the issue from statistical perspective.
> > Calculate number of records, average duration, and standard deviation
> > with averaging
> > over second/minute/hour - whatever is appropriate.
>
> > For example, for second
>
> > select
> > to_char(date_start,'YYYY-MM-DD HH24:MI:SS') d_start,
> > count(*) cnt,
> > avg((date_end - date_start)*24*3600) avg_duration
> > from tableA
> > group by
> > to_char(date_start,'YYYY-MM-DD HH24:MI:SS');
>
> Hi this makes sense, but for us to identify periods of no activity, we
> would want to get a row for every minute if the day (obviosuly for
> that purpose we would start by altering the date format masks to not
> include the :SS) - can't think at the moment how we could do that -
> any thoughts?
>
> --
> jeremy- Hide quoted text -
>
> - Show quoted text -

You may enjoy http://download.oracle.com/docs/cd/B19306_01/server.102/b14223/analysis.htm#DWHSG02013 about "Data Densification" and
http://download.oracle.com/docs/cd/B19306_01/server.102/b14223/analysis.htm#sthref1845 on "Time Series Calculations on Densified Data". You can create a densely filled matrix along e.g. the time axis and then do analysis on that. Received on Fri Oct 10 2008 - 10:03:20 CDT