udo_sas Tracker

Re: How does Box Cox parameters work?

udo_sas — Wed, 14 Jun 2017 16:53:52 GMT

Hello -

You will find some details here:

http://support.sas.com/documentation/cdl/en/etsug/66840/HTML/default/viewer.htm#etsug_tffordet_sect008.htm

Thanks,

Udo

Re: Forecast Studio: Forecasting view vs Modeling view

udo_sas — Thu, 25 May 2017 18:59:22 GMT

Hello -

This may happen if you are using a hierarchy and reconciliation.

In the modeling view you see information before your hierarchy is reconciled. Basically it shows you which models were considered and which model was chosen.

In the forecasting view you will see the impact of running reconciliation - which basically makes sure that the numbers add up across a hierarchy.

You can overlay the forecasting model plot by selecting "View -> Edit Forecasting View Properties" and check "Selected model":

This will allow you to visualize the impact of reconciliation.

Thanks,

Udo

Re: Forecasting model introduction?

udo_sas — Wed, 24 May 2017 12:00:21 GMT

Hello -

For a free online resource you may want to consult: https://www.otexts.org/fpp/

Thanks,

Udo

Re: how to import data in Forecast Studio?

udo_sas — Fri, 19 May 2017 11:51:12 GMT

Hello -

You may have set up authorizations for your libraries incorrect.

If you consult Chapter 6 of the Administrator’s Guide: "Creating and Configuring Libraries Understanding" and check out "Libraries Assigned by SAS Forecast Server" you will find details on how to go about this task.

If still in doubt my colleagues in Technical Support will be delighted to assist.

Thanks,

Udo

Re: how to import data in Forecast Studio?

udo_sas — Fri, 12 May 2017 12:45:43 GMT

Hello -

Here are four main sources of external library assignments:

All SAS libraries that are shipped with SAS are automatically assigned to each SAS session. Examples of Base SAS libraries are Sashelp, Sasuser, and Work.
Additional libraries can be assigned by adding LIBNAME statements to your SAS or server configuration files. The scope of the configuration file (in other words, is it a SAS or server configuration file) determines the availability of the library.
Libraries can also be assigned by including LIBNAME statements in the start-up code for an environment. These libraries are available only for the associated environment. If you assign a library in your environment’s start-up code, you must include a corresponding LIBNAME CLEAR statement in the shutdown code.
Pre-assigned libraries that are defined in the metadata and associated with a given server are automatically assigned when the server session is created.

More details can be found in the SAS® Forecast Server 14.2: Administrator’s Guide - Chapter 6: Creating and Configuring Libraries

Thanks,

Udo

Re: Setting threshold for forecast

udo_sas — Wed, 12 Apr 2017 12:20:30 GMT

Hello -

You may want to look into using "adjustment variables" - from SAS Forecast Studio 14.2 User's Guide (chapter 6):

"What Is an Adjustment Variable?

Systematic variations and deterministic components are included in time series data. By adjusting the time series for known systematic variations or deterministic components, the underlying stochastic (unknown) time series process can be more readily identified and modeled. For more information about adjustment variables, see the "Adjustment Step" topic in the SAS Forecast Server Procedures: User’s Guide."

Thanks,

Udo

Re: How to forecast the new product?

udo_sas — Wed, 05 Apr 2017 13:18:08 GMT

Hello -
This book provides a good overview on common approaches:
https://g.co/kgs/2243UZ

However, in my mind the approach outlines in this white paper: https://www.sas.com/content/dam/SAS/en_us/doc/whitepaper1/combining-analytics-structured-judgment-107393.pdf works well, keeping in mind that forecasting new products is most challenging.

Thanks,
Udo

Re: Amihud (2002) Liquidity Measure

udo_sas — Wed, 22 Mar 2017 21:23:00 GMT

Hello -

If you have access to SAS/ETS software you may want to use the TIMESERIES procedure - in particular the OUTSUM data set should be of interest.

Example:

proc timeseries data=sashelp.citiday outsum=work.outsum out=_null_;
id date interval=day accumulate=total;
var SNYDJCM;
run;

Thanks,

Udo

Re: Proc Expand - Convert statement

udo_sas — Tue, 28 Feb 2017 21:18:53 GMT

Hello -

If I understand your question correctly you are mostly wondering about the effect of "lead 10", correct?

proc expand data=random out=want;
id date;
convert x = x_movSumWRONG /transformout = (lead 10 reverse movesum 10 reverse); /*suggest via Tom*/
run;

Your data (random) looks like this - leaving out some data in the middle:

date	x
1-Jan-16	114.68
2-Jan-16	83.68
3-Jan-16	111.5
4-Jan-16	86.82
5-Jan-16	103.46
6-Jan-16	92.85
7-Jan-16	88.22
8-Jan-16	98.48
9-Jan-16	99.5
10-Jan-16	132.27
11-Jan-16	87.05
12-Jan-16	99.98
13-Jan-16	75.69
…	…
20-Dec-16	93.67
21-Dec-16	114.5
22-Dec-16	94.1
23-Dec-16	96.01
24-Dec-16	104.5
25-Dec-16	104.69
26-Dec-16	116.58
27-Dec-16	98.41
28-Dec-16	103.41
29-Dec-16	110.42
30-Dec-16	90.44
31-Dec-16	88.62

Lead 10 will tell EXPAND to forget about the first 10 observations (obs) - so your time series now starts at 87.05 - you will introduce missing values are the end of your series (unless you would add SETMISS 0, which would append 0s instead).
Reverse will tell EXPAND to reverse the order, the missing values you introduced will be your new first obs, 87.05 your new last obs.
Movesum 10 will create the backward moving sum of 10 obs.
Finally Reserve will put the series into its original order again.

If you decompose your "wrong" statement into:

proc expand data=random out=want;
    id date;
    convert x = x_lead10 /transformout = (lead 10);
   convert x = x_reverse /transformout = (lead 10 reverse);
   convert x = x_movsum10 /transformout = (lead 10 reverse movsum 10);
    convert x = x_reverseII /transformout = (lead 10 reverse movsum 10 reverse);
run;

you will see these columns in WANT.

In my mind this is the expected behaviour - it is just not the same what you'd like to archive with:

convert x= x_movSum /transformout = (movsum 10 trimleft 9);

Hope this makes sense,

Udo

Re: Time Series Forecasting: Holt Winter Model(Additive)

udo_sas — Mon, 20 Feb 2017 14:18:31 GMT

Hello -

You may want to have a look at: http://go.documentation.sas.com/?docsetId=etsug&docsetVersion=14.2&docsetTarget=etsug_tffordet_sect016.htm&locale=en#etsug_tffordet000750

Details are provided here: http://go.documentation.sas.com/?docsetId=etsug&docsetVersion=14.2&docsetTarget=etsug_tffordet_sect021.htm&locale=en#etsug_tffordet001867

Thanks,

Udo

Re: Testing High Frequency Seasonal Time Series for Stationarity

udo_sas — Thu, 26 Jan 2017 15:21:45 GMT

Hello -

Not sure if this paper provides you with what you are looking for: http://analytics.ncsu.edu/sesug/2010/SDA12.Dickey.pdf, but it might point you in the correct direction.

Thanks,

Udo

Re: FORECAST BASED ON HOURLY DATA

udo_sas — Mon, 05 Dec 2016 22:30:33 GMT

Hello -

It may be worthwhile to run PROC ESM instead on PROC FORECAST.

Here is an example which should work for your data (even when adding several stores):

proc esm data=production plot=forecasts lead=&lead outfor=work.AFORECASTOUTPUT;

forecast sales/model=damptrend;

by store;

run;

Of course ESM will assume the same model for all stores (in my code I used a damped trend exponential smoothing model, which seems to be appropriate for your example), but at least the parameters for each store will be different.

Thanks,

Udo

Re: How do I Forecast? Understanding, panel data, arima

udo_sas — Fri, 02 Dec 2016 18:58:42 GMT

Hello -

Please excuse for delay in responding.

1. What is the logic with BACK option?

Please check out: https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/out-of-sample-range-and-holdout-sample/m-p/225886/highlight/true#M1387 which should give you some hints.

2. Currently ESM does not allow you to incorporate inputs like logprice. My suggestion would be to look at UCM instead. Check out: http://support.sas.com/documentation/cdl/en/etsug/68148/HTML/default/viewer.htm#etsug_ucm_examples04.htm to get started.

3. You will find some additional books here:

https://www.sas.com/store/books/products-solutions/sas-ets/cBooks-cbooks_productsandsolutions-cbooks_productsandsolutions_ETS-p1.html

I would also recommend to check out: https://www.otexts.org/fpp

Thanks,

Udo

Re: How do I Forecast? Understanding, panel data, arima

udo_sas — Mon, 28 Nov 2016 23:21:33 GMT

Hello -

Not sure if my response will be useful, but you may want to double check your PROC REG code.

None of the variables used in your "MODEL" statement: model return_f=like1 omx /; seem to be part of your sample data set as far as I can tell. This is probably the reason why your PROC REG code is failing.

You wrote: "Where/how is the out sheet made and what is needed for making this forecast?"

On a very high level the flow should be as such:

a) in step 1 you create a table called "a"

b) in step 2 you will want to use this table in PROC REG - make sure that your model statement only contains variables "a" features

c) in step 2 your OUTPUT statement of PROC REG creates a table called WORK.Reg_stats

d) in step 3 you are using table WORK.Reg_stats in PROC SGPLOT to create a plot

If you get this code to run, you may want to think about if a linear regression model is indeed the best model to run for time series data.

Here is a simple forecast using ESM, which you may want to use as a reference. Instead of importing your Excel sheet I'm replicating your data in a data step and then use PROC ESM to create a forecast for the next 7 days. Note that I'm using the BACK option to compare these forecasts to data which was not used for modeling.

Hope this gives you a jump start.

Thanks,

Udo

data have ;

FORMAT Close BEST12.

Date DATE9.

;

INFORMAT Close BEST11.

Date ANYDTDTE9.

;

INPUT Close

Date

;

cards;

107.955733 8/22/2016

108.293993 8/23/2016

107.478182 8/24/2016

107.020532 8/25/2016

106.393753 8/26/2016

106.274363 8/29/2016

105.458552 8/30/2016

105.558040 8/31/2016

106.184827 9/1/2016

107.179719 9/2/2016

107.149865 9/6/2016

107.806498 9/7/2016

104.981001 9/8/2016

102.603209 9/9/2016

104.901415 9/12/2016

107.398588 9/13/2016

111.199076 9/14/2016

114.979668 9/15/2016

114.332987 9/16/2016

112.999835 9/19/2016

112.989884 9/20/2016

112.969990 9/21/2016

114.034524 9/22/2016

112.134277 9/23/2016

112.303406 9/26/2016

112.512333 9/27/2016

113.367940 9/28/2016

111.606985 9/29/2016

112.472544 9/30/2016

111.945245 10/3/2016

112.422796 10/4/2016

112.472544 10/5/2016

113.308249 10/6/2016

113.477379 10/7/2016

115.457220 10/10/2016

115.705943 10/11/2016

116.740624 10/12/2016

116.382470 10/13/2016

117.029143 10/14/2016

116.949558 10/17/2016

116.869965 10/18/2016

116.521754 10/19/2016

116.462055 10/20/2016

116.004406 10/21/2016

117.049045 10/24/2016

117.645979 10/25/2016

114.999563 10/26/2016

113.895240 10/27/2016

113.139120 10/28/2016

112.960039 10/31/2016

110.920507 11/1/2016

111.019995 11/2/2016

109.830002 11/3/2016

108.839996 11/4/2016

110.410004 11/7/2016

111.059998 11/8/2016

110.879997 11/9/2016

107.790001 11/10/2016

108.430000 11/11/2016

105.709999 11/14/2016

;

run;

proc esm data=have plot=forecasts back=7 lead=7;

id date interval=weekday accumulate=total;

forecast close / model=damptrend;

run;

Re: proc ucm intervention variables reduces r squared value

udo_sas — Mon, 21 Nov 2016 13:24:58 GMT

Hello -

One reason might be that using R Square for forecasting models is not a good idea in the first place.

See for example: http://www.forecastingprinciples.com/index.php/2013-01-30-11-08-58/10-practitioner/238-rules-for-cheaters-how-to-get-a-high-r2

Thanks,

Udo

Re: Newbie question : proc esm

udo_sas — Thu, 17 Nov 2016 13:00:54 GMT

Hello -

Little known feature of ESM - from documentation: " If an ID statement is not specified, the observation number, with respect to the BY group, is used as the time ID.". With other words, this code will work too:

proc esm data=ts.Tax

outfor=ts.outTax outstat=ts.statTax

back=0 lead=5 print=all plot=all;

forecast Million / model=damptrend ;

run;

I took the liberty to change the default model (simple ESM) to a damp trend model, which seems to be more appropriate for your data.

Thanks,

Udo

Re: From daily to monthly

udo_sas — Mon, 17 Oct 2016 08:17:58 GMT

Hello -

You may want to consider using PROC TIMESERIES instead.

For example:

proc timeseries data=work.risk out=work.want;

id date interval=month accumulate=mean;

var _numeric_;

run;

Thanks,
Udo

Re: Simple analyses on unevenly-spaced timeseries data without converting to even spacing

udo_sas — Fri, 09 Sep 2016 20:49:32 GMT

Hello -

SSM mastermind @rselukar point out to me that unevenly spaced time series are also called longitudinal data - there are 3 examples in PROC SSM documentation which deal with this type of situation:

Furthermore he was kind enough to point me to PROC SSM documentation, that describes the types of sequential data SSM can handle:

http://support.sas.com/documentation/cdl/en/etsug/68148/HTML/default/viewer.htm#etsug_ssm_details02.htm

And ouf course you may want to check out his SAS Global Forum paper: http://support.sas.com/resources/papers/proceedings15/SAS1580-2015.pdf on this very topic.

Many thanks Rajesh!

Udo

Re: Arima Forecast Evaluation

udo_sas — Thu, 25 Aug 2016 20:58:18 GMT

Hello -

http://faculty.smu.edu/tfomby/eco5375/data/Exercises/Diebold-Mariano%20Test.sas (see also: http://faculty.smu.edu/tfomby/eco5375/data/Exercises/)

might get you started.

Thanks,

Udo

Re: proc ARIMA: How to restrict forcast values to non negative

udo_sas — Tue, 02 Aug 2016 12:57:47 GMT

Hello -

As a point for consideration when going for a log transformation approach: given the nature of the exponential function the back-transformed forecasts could be behaving unexpectedly.

Thanks,

Udo