438
SQL Server 2012 Upgrade Technical Guide
Figure 12 uses SSAS 2005 to show historical forecasts (the dotted lines before the
current point in time) for the R-250 model for sales amount in Europe. What you can
see is that the forecasts are very unstable and, thus, not very reliable. You can also see
that the forecasts (the dotted lines after the current time point) become even negative
after a future time point (about 20 points in the future in this example).
Figure 12: Historical and future forecasts in SSAS 2005
The reason for this instability is that SSAS 2005 Time Series use a single algorithm,
Auto-Regression Trees with Cross-Prediction (ARTXP); this algorithm provides good
short-term forecasts only. SSAS notes this instability in long-term forecasts and simply
stops forecasting.
In SSAS 2008, 2008 R2, and 2012, you can use a blend of two different Time Series
algorithms for forecasting. Besides ARTXP, SSAS 2012 provides the Auto-Regressive
Integrated Moving Average (ARIMA) algorithm, which is much better for long-term
forecasts. After you upgrade your Time Series models to SSAS 2012, you should refine
the blend of ARTXP and ARIMA in your models by changing the FORECAST_METHOD
and PREDICTION_SMOOTHING algorithm parameters. The first parameter uses an
automatic method to determine the mixture of the algorithms. The second one
(available only in Enterprise Edition) lets you define the blend manually.
439
SQL Server 2012 Upgrade Technical Guide
As you can see in Figure 13, the upgraded version of the Time Series algorithm uses a
MIXED forecast method (default). Therefore, ARTXP is used for short-term forecasts and
ARIMA for long-term forecasts.
Figure 13: Time Series algorithm parameters in SSAS 2012
Figure 14 shows the forecast for the R-250 model for sales amount in Europe. As you
can see, forecasts quickly stabilize and even long-term forecasts never achieve
impossible values, such as values lower than zero. However, it appears that the
historical forecasts are unstable. This is because we used only forecasts for two points
in the past (the HISTORICAL_MODEL_GAP parameter), and thus only ARTXP method
was used.
440
SQL Server 2012 Upgrade Technical Guide
Figure 14: Historical and future forecasts in SSAS 2012
To learn more about Time Series algorithm parameters, see
Microsoft Time Series
Algorithm
(http://msdn.microsoft.com/en-us/library/ms174923(SQL.110).aspx) in SQL
Server 2012 Books Online.
Upgrading from SQL Server 2008 and 2008 R2
There is not much to say for upgrading data mining models from SQL Server 2008 and
2008 R2 to SQL Server 2012 for two reasons. First, data mining is a mature feature in
SQL Server. Second, there aren’t any new features in data mining in SSAS 2012, mainly
because the focus is on the Tabular mode.
Migration from SSAS 2008 and 2008 R2 to SSAS 2012 should succeed without
problems and without the need for any additional post-upgrade tasks. This is valid for
any kind of upgrade you use: in-place or side-by-side. In addition, you can use any of
the following methods for a side-by-side upgrade:
You can back up the SSAS 2008 or 2008 R2 database and restore it on SSAS
2012.
441
SQL Server 2012 Upgrade Technical Guide
With SSMS, you can create an XMLA script for creating the complete database
or any object in the database and then execute the script on SSAS 2012.
You can open the SSAS 2008 or 2008 R2 project in SSDT 2012 and deploy it on
SSAS 2012.
You can reverse-engineer an SSAS 2008 or 2008 R2 database in SSDT 2012 to
create a 2012 project and then deploy the project on SSAS 2012.
For example, Figure 15 shows the Backup Database window started from SSMS 2008 R2
in order to back up the 2008 R2 version of the database.
Figure 15: Backing up an SSAS 2008 R2 database
In Figure 16, you can see the restore window started in SSMS 2012.
442
SQL Server 2012 Upgrade Technical Guide
Figure 16: Restoring SSAS 2008 R2 database on SSAS 2012
Conclusion
There are many good reasons to upgrade your data mining models to SQL Server 2012.
If you are using SSAS 2005, you probably already measure the accuracy of your
predictive models, but you might decide to deploy a different model based on
reliability. In addition, you can get much better long-term forecasting with the Time
Series algorithm in SSAS 2012. Finally, you can consolidate multiple mining structures
into one if you need to compare mining models trained on only a subset of the
structure data. If you are upgrading from SSAS 2008 or 2008 R2, you do not gain any
new data mining features. However, you will probably want to consolidate all SSAS
databases on a single version, so upgrading your data mining models makes sense.
For upgrading your data mining models, a side-by-side migration is preferred to an in-
place upgrade. The most important reason is that with a side-by-side installation, you
leave your original models intact. However, if you do not have enough hardware power,
you can perform an in-place upgrade. With thorough testing and planning, your
upgrade can go smoothly whether your mining models are in SSAS 2005, 2008, or 2008
R2.
Dostları ilə paylaş: |