cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Search instead for 
Did you mean: 
udo_sas
Thursday
  • 417 Posts
  • 27 Likes Given
  • 43 Solutions
  • 197 Likes Received

Re: Cross sectional time series regression

by SAS Employee in SAS Forecasting and Econometrics
‎05-14-2013 12:35 PM
3 Likes
‎05-14-2013 12:35 PM
3 Likes
Hello - Some thoughts which you might find beneficial. The first thing I would do is to fill in the gaps of your data (you are facing missing values both in the time dimension and your measurements). One way to go about this is to use PROC TIMESERIES of SAS/ETS: proc sort data=have2;by ticker qtr;run; proc timeseries data=have2 out=have3 outsum=outsum; id qtr interval=qtr start="01OCT80"d end="01APR1982"d; var fraction; by ticker; run; Note that I extended all series to start and end at the same time point. OUTSUM will provide you with some statistics about your data - in particular you will find which series contain too many missing values or other problems. I would suggest to have a look at the _status_ variable first - if is is not 0, there is a problem with a particular series - see http://support.sas.com/documentation/cdl/en/etsug/65545/HTML/default/viewer.htm#etsug_timeseries_details39.htm for details on the _status_ values. You are facing about 2000 series - typically in time series forecasting assuming the same model formulation for each time series in advance (such as AR(1)) is not a good practice. Actually you may want to use a time series model formulation which account for your data pattern at hand. As you may know SAS offers software to accomplish such task in an almost automated manner - I'd suggest to check out http://www.sas.com/technologies/analytics/forecasting/index.html. Alternatively you could consider using the existing Time Series Forecast System in SAS/ETS (see http://support.sas.com/documentation/cdl/en/etsug/63939/HTML/default/viewer.htm#tfintro_toc.htm) but note that it does not support any BY variables - you will need to create a new data mart which features one variable per series in your original data set. Finally: for creating lagged variables you may want to have a look at the LAG function - see: http://support.sas.com/documentation/cdl//en/lefunctionsref/63354/HTML/default/viewer.htm#n0l66p5oqex1f2n1quuopdvtcjqb.htm Hope this is useful. Thanks, Udo ... View more

Re: Cross sectional time series regression

by SAS Employee in SAS Forecasting and Econometrics
‎05-13-2013 04:51 PM
‎05-13-2013 04:51 PM
Hello - If you have access to SAS/ETS software, you should go ahead and have a look at either: The ARIMA procedure provides the identification, parameter estimation, and forecasting of autoregressive integrated moving-average (Box-Jenkins) models, seasonal ARIMA models, transfer function models, and intervention models The AUTOREG procedure provides regression analysis and forecasting of linear models with autocorrelated or heteroscedastic errors I looked at your data, but I got confused about your usage of "prevFraction". Initially I thought that prevFraction is the lagged value of Fraction, but it is not. For example for ticker AAPL: For other tickers there are data quality problems like different length of history, missing information, etc (see ABLE or ADC for example). If prevFraction does not feature lagged values of Fraction, then your equation above does not work - you will have to come up with an ARIMAX formulation for example, which features prevFraction as an input variable. If prevFraction is indeed a lag1 value of Fraction, then you don't need to specify a column - you can use standard ARIMA syntax. Thanks, Udo ... View more

Re: Exponential Smoothing Weights and Automatic Forecasting

by SAS Employee in SAS Forecasting and Econometrics
‎05-13-2013 02:11 PM
1 Like
‎05-13-2013 02:11 PM
1 Like
Hello - When using the automatic diagnostic option in SAS Forecast Studio for ESM the smoothing weights are restricted between 0.0001 and 0.9999. In order to apply different weights settings you can: go to "Modeling View" - create a copy of the system generated ESM - and edit your copy. Then you should see the following dialog: Under Weights you can change the restrictions. Parameters and their restrictions are required to be greater than or equal to –1 and less than or equal to 2. In order to extend the automatic behavior of SAS Forecast Studio you can add you own ESM repository under "models from an external list" (Model Generation option in Forecast settings). You will need to specify your ESMs using the HPFESMSPEC and HPFSELECT procedures. Hope this makes sense, Udo ... View more

Re: White Noise inputs in forecasting ARMA variables

by SAS Employee in SAS Forecasting and Econometrics
‎05-10-2013 04:50 PM
‎05-10-2013 04:50 PM
Hello - Even if you are not using PROC ARIMA for estimation, you may want to check out http://support.sas.com/documentation/cdl/en/etsug/63939/HTML/default/viewer.htm#etsug_arima_sect015.htm and http://support.sas.com/documentation/cdl/en/etsug/63939/HTML/default/viewer.htm#etsug_arima_sect042.htm for some ideas. Thanks, Udo ... View more

Re: PROC HPFTEMPRECON -- week to month reconciliation?

by SAS Employee in SAS Forecasting and Econometrics
‎05-07-2013 01:12 PM
‎05-07-2013 01:12 PM
Hello - This is an example for illustration purposes only - using your data using temporal reconciliation might turn out to be more useful. The challenge is to get the boundaries of the different aggregation levels right (otherwise you might end up with not fully observed intervals). As you were wondering about using custom intervals: HPFTEMPRECON currently does not suggest custom intervals. Thanks, Udo *create sample data; data subset(keep=date volume);   set sashelp.timedata;   date=datepart(datetime);   if date lt "01JAN1998"d or date gt "31DEC99"d then delete;   format date date9.; run; proc timeseries data=subset out=subset;   id date interval=week accumulate=total; var volume; run; *forecast on weekly frequency; proc hpfengine data=subset rep=sashelp.hpfdflt globalselection=tsfsselect out=_null_ outfor=outforwk plot=forecastslead=5;   id date interval=week accumulate=total;   forecast volume; run; *forecast on monthly frequency - using R445; proc hpfengine data=subset rep=sashelp.hpfdflt globalselection=tsfsselect out=_null_ outfor=outformon plot=forecastslead=1;   id date interval=R445MON accumulate=total;   forecast volume; run; *convert weekly predictions to daily; proc expand data=outforwk(keep=_name_ actual predict date) out=outforday from=week to=day;   id date; convert predict; by _name_; run; *temporal reconcilation of month to day; proc hpftemprecon data=outforday benchdata=outformon outfor=benfor; id date interval=day; benchid date interval=R445MON; run; *aggregate adjusted daily forecast to weekly frequency; proc timeseries data=benfor out=series; id date interval=week accumulate=total; var predict; run; *merge results of weekly forecasts and adjusted weekly forecasts and plot; data result; merge outforwk series(rename=(predict=bench)); by date; run; proc sgplot data=result;   series x=date y=actual;   series x=date y=predict;   series x=date y=bench; run; ... View more

Re: PROC HPFTEMPRECON -- week to month reconciliation?

by SAS Employee in SAS Forecasting and Econometrics
‎05-06-2013 01:31 PM
‎05-06-2013 01:31 PM
Hello - This paper: http://support.sas.com/resources/papers/proceedings11/326-2011.pdf provides more details on using HPFTEMPRECON and its limitations. In particular the authors point out that weeks don't have to be fully nested in months, as a week can span two months. Therefore, the frequency of the indicator series cannot be weekly when the benchmark series has a monthly frequency. One strategy for overcoming this issue for point forecasts (i.e. PREDICT) is as follows: Use PROC EXPAND to generate daily forecasts from the weekly forecasts Use PROC HPFTEMPRECON to reconcile the daily forecasts to the monthly forecasts Accumulate the reconciled daily forecasts to the weekly interval using PROC TIMESERIES Hope this is useful. Thanks, Udo ... View more

Re: Rolling Simulations in Forecast Studio

by SAS Employee in SAS Forecasting and Econometrics
‎05-02-2013 11:26 AM
2 Likes
‎05-02-2013 11:26 AM
2 Likes
Hello - There is a conference paper on rolling simulations which was not published yet - I will send it to you directly. Thanks, Udo ... View more

Re: PROC UCM estimates

by SAS Employee in SAS Forecasting and Econometrics
‎04-11-2013 04:27 PM
‎04-11-2013 04:27 PM
Hello - Would it be possible to share your code and some example data? As a wild guess: If you are using a local linear model, you could try to set the LEVEL variance to 0, but leave the SLOPE variance to be determined by the procedure. This will result in a smoother trend estimation, which might be helpful. Thanks, Udo ... View more

Re: Times Series using Enterprise guide: problem

by SAS Employee in SAS Forecasting and Econometrics
‎04-03-2013 08:43 AM
‎04-03-2013 08:43 AM
Hello - Can you share some more details on the problem you are facing with HPF? Ideally sharing a code example and test data for illustration purposes... Thanks, Udo ... View more

Re: Proc Statespace Procedure

by SAS Employee in SAS Forecasting and Econometrics
‎03-20-2013 08:44 AM
‎03-20-2013 08:44 AM
Hello - There are various forms of the state space model in use. The form of the state space model used by the STATESPACE procedure is based on Akaike and covers a specific formulation only. The UCM procedure (http://support.sas.com/documentation/cdl/en/etsug/63939/HTML/default/viewer.htm#ucm_toc.htm) provides access to a more flexible approach, but is restricted to one dependent variable. The new SSM procedure (http://support.sas.com/documentation/cdl/en/etsug/63939/HTML/default/viewer.htm#ssm_toc.htm - experimental with 9.3) provides access to the broadest range of state space models (including multiple dependent) and extents the UCM procedure. This new procedure is a much more powerful and general tool for state space modeling than the old STATESPACE procedure, in fact the STATESPACE procedure should now be considered somewhat obsolete and superseded by SSM. You will find examples for UCM and SSM syntax in the respective chapters of documentation. Hope that helps, Udo ... View more

Re: Rid off from error message in proc forecast.

by SAS Employee in SAS Forecasting and Econometrics
‎03-18-2013 03:38 PM
3 Likes
‎03-18-2013 03:38 PM
3 Likes
Hello - You may try the following: data pred;       a=1;b=1;output;       a=1;b=2;output;       a=1;b=3;output;       a=1;b=4;output;       a=1;b=5;output;       a=2;b=2;output;       a=2;b=4;output; run; proc esm data=pred lead=1 out=next;       forecast b / method=linear;       by a; run; A couple of notes: ESM runs exponential smoothing models with optimized parameters. Given your example a linear (Holt) exponential smoothing model seems to make sense, but the question remains if this is always the case (is a linear ESM always appropriate?). The FORECAST procedure will apply the STEPAR Algorithm by default - its data requirements are described here: http://support.sas.com/documentation/cdl/en/etsug/63939/HTML/default/viewer.htm#etsug_forecast_sect023.htm If you have only a few observations (like 2 in your example) available any forecasting technique becomes questionable of course. Maybe you are more concerned about fitting a linear regression line to your data and then extrapolate it to include new observations? In this case you are better off using the REG procedure of course. The question is if your data is autocorrelated. Thanks, Udo ... View more

Re: explanatory variables in garch variance equation

by SAS Employee in SAS Forecasting and Econometrics
‎03-14-2013 09:10 AM
‎03-14-2013 09:10 AM
Hello - This might be a Technical Support issue - difficult to judge with the current amount of information you shared. Would it be possible to share you data as well? Alternatively you might consider contacting Technical Support. Thanks, Udo ... View more

Re: Incomplete time series how to deal with them?

by SAS Employee in SAS Forecasting and Econometrics
‎03-13-2013 08:56 AM
‎03-13-2013 08:56 AM
Hello - All time series based forecasting methods have a similar requirement for the ID variable (date or datetime information): it needs to be complete, with other words missing values in the time ID are not allowed. As such you will have to fill in these gaps for the time ID - for example by using PROC TIMESERIES: data have; set sashelp.air; if date="01JAN57"d then delete; if date="01SEP59"d then delete; run; proc timeseries data=have out=want; id date interval=month accumulate=total; var air; run; One the time id is intact, for UCM the following holds true (from documentation): "Embedded missing values in the dependent variable usually cause no problems in UCM modeling. However, no embedded missing values are allowed in the predictor variables." Depending on your business question at hand you may want to impute missing values of the dependent varialbe using PROC TIMESERIES for example (see setmissing option). Thanks, Udo ... View more

Re: Use proc expand for specific cases

by SAS Employee in SAS Forecasting and Econometrics
‎03-11-2013 02:54 PM
‎03-11-2013 02:54 PM
Hello - I will give it a try, but I'm pretty sure I'm missing the mark - in this example I'm using EXPAND to "smooth a pattern" with a moving average window of 5 observations: proc expand data=sashelp.air out=out method=none;      id date;      convert air=predict / transformout=(movave 5); run; data plot; set sashelp.air out; by date; run; proc sgplot data=out; series x=date y=air; series x=date y=predict; run; It would be very helpful if you could an example (how does the data look like now? How would you like the results to look like?). Thanks, Udo ... View more

Re: Books for churn , propensity modeling using SAS, Clustering techni...

by SAS Employee in SAS Forecasting and Econometrics
‎03-01-2013 03:54 PM
‎03-01-2013 03:54 PM
Hello - You may want to post re-post in the data mining community: https://communities.sas.com/community/support-communities/sas_data_mining_and_text_mining Thanks, Udo ... View more
Contact Me
Online Status
Offline
Date Last Visited
Thursday
Likes given to
Likes from
Quick links