Date: Thu, 25 Sep 2003 22:48:57 +0800
Reply-To: Tans Family <tans_family@PACIFIC.NET.SG>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Tans Family <tans_family@PACIFIC.NET.SG>
Subject: Re: Calculating Average Transactional Lapse between dates
Content-Type: text/plain; charset="iso-8859-1"
Hi Martins,
Thanks for your reply.
No it is not using the last trans date to substract the first trans date.
It is really going into each and every transaction to get the differences
and add together to divide by the transaction gaps.
---------------------------------------------------------------------------
Example:
-----------------------
Member Date difference_in_days
1 08/30/03
1 09/30/03
1 12/20/03
1 12/30/03
Therefore it should be
1 08/30/03
1 09/30/03 31days (30sep - 30aug)
1 12/20/03 82days (20Dec - 30Sep)
1 12/30/03 10days (30Dec - 20Dec)
--------------------------------------------
Compute avg_trans_lapse = (31 + 82 + 10) / 3
Thanks very much for helping me!!!
Appreciate!
----- Original Message -----
From: <Martins.Liberts@csb.gov.lv>
To: "Tans Family" <tans_family@PACIFIC.NET.SG>
Cc: <SPSSX-L@LISTSERV.UGA.EDU>
Sent: Thursday, September 25, 2003 9:38 PM
Subject: Re: Calculating Average Transactional Lapse between dates
>
> Hi Tans,
>
> Is [(30Sep - 30 Aug) + (20Dec - 30Sep)] the same as [20Dec - 30 Aug]?
>
> If so then try this:
>
> agg outfile=*
> /break=member
> /trd_s=min(trandate)
> /trd_f=max(trandate)
> /n=N.
>
> comp antd=(trd_f-trd_s)/(60*60*24)/(n-1).
> exe.
>
>
> --
> Martins Liberts
> Central Statistical Bureau of Latvia
> address: Lacplesa iela 1, Riga, LV-1301, Latvia
> phone: 371-7366877
> email: Martins.Liberts@csb.gov.lv
>
>
>
> Tans Family
> <tans_family@PACIF To:
SPSSX-L@LISTSERV.UGA.EDU
> IC.NET.SG> cc:
> Sent by: "SPSSX(r) Subject: Calculating
Average Transactional Lapse between dates
> Discussion"
> <SPSSX-L@LISTSERV.
> UGA.EDU>
>
>
> 25.09.2003 12:09
> Please respond to
> Tans Family
>
>
>
>
>
>
> Dear List,
>
> data list list/member(f2.0) trandate (adate9).
> begin data.
> 1 08/30/03
> 1 09/30/03
> 1 12/20/03
> 5 02/15/03
> 5 10/18/03
> 5 11/25/03
> 5 12/22/03
> end data.
>
> I have a transactional data set which I would like to compute the average
> no. of transactional days between each transaction record.
>
> Is this possible?
>
> Example:
> --------------------------
>
> For member 1,
>
> His transactional lapse should be [(30Sep - 30 Aug) + (20Dec - 30Sep)]
> Divided by 2
> -----------------------------
> Please advise.
>
> Thank you
> EC
>
>
>
>