About podarum

podarum · ‎03-28-2012

Hi, I realize the subject line may confuse many, but here's what I'm wondering if someone can help. I have over 1500 time series (stores by products) that have different end dates. For example one store may have thier Price data from Jan'1996 to Jun'2010 and another from March'2000 to Dec'2011, and so on for each different Sore and Product. I also do have other variables other than Price that also ends differently by month and year for each Store x Product (since it's from many different sources). Is there a way in Proc Timeseries or any other procedure that I can compute (in a sense forecast) all data to have a similar end-date for each StorexProduct (for example Dec'2011). I eventually want to use ARIMAX to forecast Prices for the next 12 months, but I have to get all series to the same ending line, or starting line depends how you look at it.... Thanks in advance.

podarum · ‎03-28-2012

What can I say.. Brilliant.. not only did it work perfectly, it was on the first shot.. great stuff, Thanks

podarum · ‎03-28-2012

For example have a field called MISSING, and if the value is missing at the start then have an 'S' in that field, and if the missing is at the end then have an 'E" in that cell, and nothing if the value is not missing or if it's missing in between.

podarum · ‎03-28-2012

Worked fine.. thanks.. now I have to get a flag for the missing

podarum · ‎03-28-2012

This is great stuff .. thanks

podarum · ‎03-27-2012

How could I find how many observations are missing at the front and end, By group.. So for example: HAVE Store Var1 Var2 Var3 A . 4 . A . 5 . A 4 . 4 A 6 . 5 A . . 6 B 4 . 9 B 5 4 . B . 6 8 It would tell me that Store A - Var1 has 2 obs missing at the front and 1 at the end Store B - Var1 has 1 at end Store A - Var2 has 3 at the end Store B - Var2 has 1 at the front Store A - Var3 has 2 at the front Store B - Var3 has NONE Even better if there is a way I can identify them through an indicator or something, that way I can delete thenm if I chose. thank you..

podarum · ‎03-23-2012

The problem is that I have 500 stores and 5 products, so a combo of 2500 time-series models.. but when I try to fill in the gaps using the above 'expand' code, I end up with 490 stores and 5 products.. something like this: Store Date Price Product A 2000.01 234 A1 A 2000.02 246 A1 A 2000.03 258 A1 B 2000.01 564 B2 B 2000.02 534 B2 B 2000.03 533 B2 B 2000.01 345 B1 B 2000.02 354 B1 C 2005.03 356 C2 C 2005.04 367 C2 C 2005.05 359 C2 .. and so on.....(there are alos many other variables attached, just didn't specify them down)..

podarum · ‎03-22-2012

Thanks FriedEgg, very interesting approach.. And I had no idea the usage of coalesce, its so usefull.

podarum · ‎03-22-2012

proc expand data=MS.TC_CMA out=MS.TC_CMA_expand to=MONTH; by Store Product Store_CD City Province; convert Price counts Index YR_Opt Value DOM / method=JOIN; id Date; format Date YYMMP7.; run;

podarum · ‎03-22-2012

Hi, If anyone knowledgeable in proc expand, plese explain to me.. I have prices by month/year (time-series) of about 500 stores and 5 products... The prices for some stores are missing. So I used proc expand to interpolate and fill in the holes.. after running the proc expand a small number of stores dissapeared.. and now left with 490 stores.. why is this happening? thanks

podarum · ‎03-21-2012

Sorry , I think I was getting 0's because I had none missing.. for a second I was thinking I was only seeing the perecentage of missing... All clear now.. thanks

podarum · ‎03-21-2012

I eyeballed the data and each of the observations has a Price listed, so it would be 100% and another product would be perhaps 90% , but am getting 0% for both..

podarum · ‎03-21-2012

I'm stumped.. I treid everything.. it all makes sense, all the other variables by themselves are fine, except this Price.. could it be that Format = COMMA9. and Informat = BEST32. ?

podarum · ‎03-21-2012

Just to make sure that I'm doing this correctly.. I have 13 varaibles that I need to convert 0's to missing.. so I used this: array v (*) Price--NHIP; do i = 1 to dim(v);if v(i) = 0 then v(i) = .;end; and then; select Store, Product, (nmiss(Price)) / Count(*)*1) as result Does this look correct ?

podarum · ‎03-21-2012

I ran the same code and got the correct results on the test data, but seem sshort on the main data.. I did change the 0's to . .. I'll try again..

Online Status	Offline
Date Last Visited	‎02-11-2026 09:56 AM

Re: How to find a string within anther string for different scenarios

How to find a string within anther string for different scenarios

Keep only part of a symput address path stationary

Update and read excel data with drop-down column in Excel

Re: Group data into specific days of the week

Group data into specific days of the week

Finding if scheduled SAS job ran successfully in Virtual Machine

See the number of rows created and if successful pull

Re: importing an xlsm file (Windows 10)

importing an xlsm file (Windows 10)

Re: How to find a string within anther string for different scenarios

Re: How to find a string within anther string for different scenarios

Re: Make row values into column

Re: Importing cuts-off field length

Re: getting week number from middle of year to next middle of year

Make first letter capital only

Exclude columns using proc SQL

Group data into specific days of the week

use date in a select range

sum up across columns and have a criteria

how to time advance many time series?

Re: find the # of observations missing before & after

find the # of observations missing before & after

find the # of observations missing before & after

find the # of observations missing before & after

find the # of observations missing before & after

proc expand deletes data ?

Re: detect % of observations missing using BY

proc expand deletes data ?

proc expand deletes data ?

Re: detect % of observations missing using BY

Re: detect % of observations missing using BY

Re: detect % of observations missing using BY

Re: detect % of observations missing using BY

Re: detect % of observations missing using BY