About the Series …
This article is a member of the series MDX in
Analysis Services. The series is designed to provide hands-on
application of the fundamentals of MDX from the perspective of MS SQL Server
2000 Analysis Services; our primary focus is the manipulation of
multidimensional data sources, using MDX expressions, in a variety of scenarios
designed to meet real-world business intelligence needs.
information on the series, as well as the hardware / software requirements to
prepare for the tutorials we will undertake, please see the first lesson of
this series: MDX Concepts and Navigation.
the time of writing, Service Pack 3 / 3a updates are assumed for MSSQL Server 2000, MSSQL
Server 2000 Analysis Services, and the related Books Online
and Samples. Images are from a Windows 2003 Server
environment, upon which I have also implemented MS Office 2003, but the
steps performed in the articles, together with the views that result, will be
quite similar within any environment that supports MSSQL Server 2000 and
MSSQL Server 2000 Analysis Services ("Analysis Services" or "MSAS").
The same is generally true, except where differences are specifically noted,
when MS Office 2000 and above are used in the environment, with respect
to any MS Office components presented in a given article.
this article, we resume the focus of a group of articles that began with Mastering
Time: Change across Periods. In that article, as well as its immediate successor, Mastering
Time: Period – to – Date Aggregations,
we concentrated upon the Time dimension from the
perspective of our MDX queries. Our intent, in these and occasional subsequent
articles, is to explore ways to effectively report change over time, as
well as to accumulate those changes to present snapshots, trends and
other time-based metrics in a precise manner to meet typical business
requirements. As most of us realize, time is the most pervasive
dimension. A cube that has no time dimension is rare, indeed. Consequently,
this group of articles holds information that is of interest to virtually
anyone involved with MSAS cube design, development and use.
this article, we will examine "rolling average" aggregations,
a common business requirement. "Rolling," or "moving,"
averages, involve a measure, the average under consideration, that is
aggregated over a progressively moving window of time periods. (While the
window typically involves time, the functions that define the "window"
involved here can certainly involve members of other dimensional levels).
Rolling averages are
popular in the business community because they have the effect of smoothing the
values of a quantity that fluctuates over time; these moving averages can be
especially useful in cases where the values to which they are applied are
subject to seasonal variations and other volatility factors.
They aid us in "normalizing," or "flattening," the presentation
of the metric for evaluation purposes. An example might be the S & P 500
Annual Yield 12-Month Rolling average, from a specific point in the past to
recent times, a representation of which is depicted in Illustration 1.
Illustration 1: S&P 500 Annual
Yield 12 Month Rolling Average %, 1947 to Present
have stated, the value of the moving average, whether presented in chart,
tabular or other reports, often lies in its capacity to free us from some of the
distraction of fluctuations that are meaningless, or at least not completely
relevant, when it comes to trying to see long-term patterns in the analysis of
a quantity / measure. Because
they smooth a data series and make it easier to spot trends (something that is
especially helpful in volatile markets, and in many other environments, as
well), moving averages are one of the most popular and easy to use tools
available to the technical analyst. Moving averages also form the building
blocks for many other technical indicators and overlays.
In addition to being applied in the realm of stock prices, rolling
averages are used with many other metrics that change frequently. We might,
for instance, create a report to display weekly sales revenue over a three-year
window. We could, in this example, plot the figures for our organization’s
sales revenue for each of the weeks, along with another row (or line, in the
case of a chart) that displays a cumulative or a multi-week rolling average. A
rolling average is generated simply by calculating the average of the current
value, together with the specified number of previous values. The
individual values are, of course, typically identified by a time period.
affords us several approaches to generating rolling aggregates. We will explore
one of these in this article, and another in the next article of this series.
In both articles, we will overview the means of managing a rolling average
requirement, using MDX within MSAS to accomplish our ends. We will then
undertake a multi-step practice example that activates the underlying concepts,
discussing our objectives, as well as the results we obtain, with each step. We will:
considerations applicable to our approach, commenting generally upon the
environment within which we will perform our practice exercises;
hypothetical business requirement, in which a group of information consumers
have requested a particular moving average capability for analysis purposes;
Use the Sample
Application to construct and test the MDX required to support a rolling average
calculated member in Analysis Manager to provide permanent rolling average
support in the cube;
accuracy of operation once again, from the Data view within the cube;
that the rolling average calculated measure behaves in a "contextually
sensitive" way, within the context of the time dimension levels.