topic Re: How to use ARIMA model to forecast a data series in SAS Forecasting and Econometrics

topic Re: How to use ARIMA model to forecast a data series in SAS Forecasting and Econometrics https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115962#M640 <HTML><HEAD></HEAD><BODY><P>Hi Udo,</P><P>Thank you very much for your great help.</P><P>Your help is very valuable for us.</P><P></P><P>Best regards</P><P></P><P>Mirisage</P><P><BR /> </P></BODY></HTML> Wed, 25 Jul 2012 21:15:05 GMT Mirisage 2012-07-25T21:15:05Z How to use ARIMA model to forecast a data series https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115959#M637 <HTML><HEAD></HEAD><BODY><P><SPAN style="font-family: 'Courier New'; background: white; color: green; font-size: 11pt;">Hi Forum,</SPAN></P><P></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green; font-size: 11pt;">I have a variable named "ninety" in the attahced file. I need to forecast its values for maximum possible length ahead </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green; font-size: 11pt;">as dictated by my 24 month's actual observations.</SPAN></P><P></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black; font-size: 11pt;">I have tried various methods such as moving average (3point, 12point), weighted moving average, exponential smoothing etc. but the Percent Absolue Deviation between predicted and actual values are way too high. </SPAN></P><P></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black; font-size: 11pt;">So, I tried ARIMA as the last resort. </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black; font-size: 11pt;">  </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black; font-size: 11pt;">Honestly I do not have any previous ARIMA expereince, so I have followed the SAS documentation mecahnically and followed the steps. </SPAN></P><P></P><P><SPAN style="font-size: 12pt; text-decoration: underline;"><STRONG style="color: #ff00ff; background: white; text-decoration: underline; font-family: 'Courier New';">Question: I cannot run the last two steps to get the final forecasts. Could someone please help?</STRONG></SPAN></P><P> </P><P><SPAN style="color: #008000; font-size: 12pt;"><STRONG style="background: white; font-family: 'Courier New';">Thanks, Mirisage</STRONG></SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green;">/*taking the graph*/</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: blue;">ods</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">graphics</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">on</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">sgplot</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test;</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: blue;">scatter</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">y</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=ninety </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: blue;">x</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=current_date;</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green;">/*Analyzing correlation properties*/</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">arima</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test ;</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: blue;">identify</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">var</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=ninety </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: blue;">nlag</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=</SPAN><STRONG style="color: teal; background: white; font-family: 'Courier New';">6</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P></P><P></P><P></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green;">/*White Noise Test*/</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">arima</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test ;</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: blue;">identify</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">var</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=ninety </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: blue;">nlag</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=</SPAN><STRONG style="color: teal; background: white; font-family: 'Courier New';">6</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">arima</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test ;</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: blue;">identify</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">var</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=ninety </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: blue;">nlag</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=</SPAN><STRONG style="color: teal; background: white; font-family: 'Courier New';">5</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">arima</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test ;</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: blue;">identify</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">var</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=ninety </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: blue;">nlag</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=</SPAN><STRONG style="color: teal; background: white; font-family: 'Courier New';">4</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">arima</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test ;</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: blue;">identify</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">var</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=ninety </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: blue;">nlag</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=</SPAN><STRONG style="color: teal; background: white; font-family: 'Courier New';">3</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green;">/*Since the series is nonstationary, I transformed it to a stationary series by differencing.*/</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green;">/*Identification of the Differenced Series*/</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">arima</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test;</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: blue;">identify</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">var</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=ninety(</SPAN><STRONG style="color: teal; background: white; font-family: 'Courier New';">1</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">);</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green;">/*The autocorrelations decrease rapidly in this plot, indicating that the </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: green;">change in ninety is a stationary time series. */</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: #ff0000; font-size: 14pt;">/*not running*/</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">arima</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: blue;">estimate</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">p</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=</SPAN><STRONG style="color: teal; background: white; font-family: 'Courier New';">1</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P>   <STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: #ff0000; font-size: 12pt;">/*Not running*/</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: green;">/*forecasting*/</SPAN></P><P><STRONG style="color: navy; background: white; font-family: 'Courier New';">proc</STRONG> <STRONG style="color: navy; background: white; font-family: 'Courier New';">arima</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">data</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=test;</SPAN></P><P>   <SPAN style="font-family: 'Courier New'; background: white; color: blue;">forecast</SPAN> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">lead</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=</SPAN><STRONG style="color: teal; background: white; font-family: 'Courier New';">12</STRONG> <SPAN style="font-family: 'Courier New'; background: white; color: blue;">interval</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=month </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: blue;">id</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=date </SPAN><SPAN style="font-family: 'Courier New'; background: white; color: blue;">out</SPAN><SPAN style="font-family: 'Courier New'; background: white; color: black;">=results;</SPAN></P><P>   <STRONG style="color: navy; background: white; font-family: 'Courier New';">run</STRONG><SPAN style="font-family: 'Courier New'; background: white; color: black;">;</SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P><P><SPAN style="font-family: 'Courier New'; background: white; color: black;"> </SPAN></P></BODY></HTML> Mon, 23 Jul 2012 18:56:25 GMT https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115959#M637 Mirisage 2012-07-23T18:56:25Z Re: How to use ARIMA model to forecast a data series https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115960#M638 <HTML><HEAD></HEAD><BODY><P>Hi Udo,</P><P></P><P>If you have time, could you please have a look on this ARIMA query. I really apprecaite.</P><P></P><P>Thank you</P><P></P><P>Mirisage</P><P></P><P><BR /> </P></BODY></HTML> Tue, 24 Jul 2012 13:49:59 GMT https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115960#M638 Mirisage 2012-07-24T13:49:59Z Re: How to use ARIMA model to forecast a data series https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115961#M639 <HTML><HEAD></HEAD><BODY><P>Update: just noticed that my ARIMA code got truncated unfortunately - corrected it now.</P><P></P><P>Hello -</P><P>You might want to combine your last 3 PROC ARIMA calls into 1 call (and provide the correct name of your ID variable in the forecast statement):</P><P><SPAN style="font-size: 10pt; font-family: Courier New;"> </SPAN><SPAN style="; color: #000080; font-size: 10pt; font-family: Courier New;"><STRONG>proc arima</STRONG></SPAN><SPAN style="font-size: 10pt; font-family: Courier New;"> </SPAN><SPAN style="color: #0000ff; font-size: 10pt; font-family: Courier New;">data</SPAN><SPAN style="font-size: 10pt; font-family: Courier New;">=test;</SPAN></P><P><SPAN style="font-size: 10pt; font-family: Courier New;"><SPAN style="color: blue; font-size: 10pt; background: white; font-family: 'Courier New';">identify</SPAN> </SPAN><SPAN style="color: #0000ff; font-size: 10pt; font-family: Courier New;">var</SPAN><SPAN style="font-size: 10pt; font-family: Courier New;">=ninety(</SPAN><SPAN style="; color: #008080; font-size: 10pt; font-family: Courier New;"><STRONG>1</STRONG></SPAN><SPAN style="font-size: 10pt; font-family: Courier New;">);</SPAN></P><P><SPAN style="font-size: 10pt; font-family: Courier New;"> </SPAN><SPAN style="color: #0000ff; font-size: 10pt; font-family: Courier New;"><SPAN style="color: blue; font-size: 10pt; background: white; font-family: 'Courier New';">estimate </SPAN>p</SPAN><SPAN style="font-size: 10pt; font-family: Courier New;">=</SPAN><SPAN style="; color: #008080; font-size: 10pt; font-family: Courier New;"><STRONG>1</STRONG></SPAN><SPAN style="font-size: 10pt; font-family: Courier New;">;</SPAN></P><P><SPAN style="background: white; color: blue; font-family: 'Courier New';">forecast</SPAN> <SPAN style="background: white; color: blue; font-family: 'Courier New';">lead</SPAN><SPAN style="background: white; color: black; font-family: 'Courier New';">=</SPAN><SPAN style="background: white; color: teal; font-family: 'Courier New';"><STRONG>12</STRONG></SPAN> <SPAN style="background: white; color: blue; font-family: 'Courier New';">interval</SPAN><SPAN style="background: white; color: black; font-family: 'Courier New';">=month </SPAN><SPAN style="background: white; color: blue; font-family: 'Courier New';">id</SPAN><SPAN style="background: white; color: black; font-family: 'Courier New';">=current_date </SPAN><SPAN style="background: white; color: blue; font-family: 'Courier New';">out</SPAN><SPAN style="background: white; color: black; font-family: 'Courier New';">=results;</SPAN></P><P><SPAN style="background: white; color: navy; font-family: 'Courier New';"><STRONG>run</STRONG></SPAN><SPAN style="background: white; color: black; font-family: 'Courier New';">;</SPAN></P><P></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';">Since you have access to SAS/ETS it seems <img id="smileyhappy" class="emoticon emoticon-smileyhappy" src="https://communities.sas.com/i/smilies/16x16_smiley-happy.png" alt="Smiley Happy" title="Smiley Happy" /> you could also feed your series to the Time Series Forecasting System:</SPAN></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';">Start SAS on your machine and take advantage of the automatic model selection. </SPAN></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';"><IMG __jive_id="2306" alt="Screenshot.PNG" class="jive-image-thumbnail jive-image" src="https://communities.sas.com/legacyfs/online/2306_Screenshot.PNG" width="450" /></SPAN></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';">Documentation about TSFS can be found here:</SPAN></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';"><A class="active_link" href="http://support.sas.com/documentation/cdl/en/etsug/60372/HTML/default/viewer.htm#tfintro_toc.htm" title="http://support.sas.com/documentation/cdl/en/etsug/60372/HTML/default/viewer.htm#tfintro_toc.htm">http://support.sas.com/documentation/cdl/en/etsug/60372/HTML/default/viewer.htm#tfintro_toc.htm</A></SPAN></P><P></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';">BTW: your series seems to feature a level shift after Nov2010 - if you can explain why this might be good modeling information:</SPAN></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';"><IMG __jive_id="2307" alt="SeriesPlot1.png" class="jive-image-thumbnail jive-image" src="https://communities.sas.com/legacyfs/online/2307_SeriesPlot1.png" width="450" /></SPAN></P><P><IMG __jive_id="2308" alt="CyclePlot1.png" class="jive-image-thumbnail jive-image" src="https://communities.sas.com/legacyfs/online/2308_CyclePlot1.png" width="450" /></P><P>To replicate the graphics above use the following code (you might want to experiment with the plot features of this ETS procedure as well):</P><P></P><P>proc timeseries data=test out=_null_ plot=(series cycles);</P><P>id current_date interval=month;</P><P>var ninety;</P><P>run;</P><P></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';">Thanks,</SPAN></P><P><SPAN style="background: white; color: black; font-family: 'Courier New';">Udo</SPAN></P></BODY></HTML> Tue, 24 Jul 2012 18:28:34 GMT https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115961#M639 udo_sas 2012-07-24T18:28:34Z Re: How to use ARIMA model to forecast a data series https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115962#M640 <HTML><HEAD></HEAD><BODY><P>Hi Udo,</P><P>Thank you very much for your great help.</P><P>Your help is very valuable for us.</P><P></P><P>Best regards</P><P></P><P>Mirisage</P><P><BR /> </P></BODY></HTML> Wed, 25 Jul 2012 21:15:05 GMT https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/How-to-use-ARIMA-model-to-forecast-a-data-series/m-p/115962#M640 Mirisage 2012-07-25T21:15:05Z