Go ahead means improve little by
little, so just now (once I attended “IBM SPSS Statistics within Windows” course
in Seville University) not only I would like to organize but also share some of my findings, in order to involve SPSS results as a complement in Oracle
Business Intelligence Dashboard.
In another words, once people keep well
informed of what’s going on (based on access to accurate
information i.e.: OBIEE), then is required deliver conclusions (make
sense of information i.e.: IBM SPSS Statistics ).
In this way, as soon as I noticed
about IBM SPSS Statistics results,suddently I wondered which one is the best way in
order to display such results within OBIEE in a single screen.
Because I realized that make sense what otherwise would have remained
hidden in the numbers (like a forest would have remained hidden in
the trees).
And out of frustation, as a result ,
conclusions are displayed within OBIEE as a single picture, by
properly enconding
information using SPSS:
So, in this post, let’s talk about
analysis of variance (ANOVA) as an useful tool focus on independence
between a response variable (variable that are measured) and factors
(variable that can be controlled).
Anyway, in this case, let me say that a
background is required, in order to get knowledge about what SPSS
delivers as a result.
In concret, we focus on one factor
analysis of variance, which involves a single factor. Each factor
includes different levels i.e. as it shows the next slide:
Once assumptions are validated (using
SPSS), then contrast can be helpful in order to determine if the null
hypotesis is accepted:
H0: α1 = α2 =
α3
H1: αi ≠ αj where
i ≠ j so not all are equal
Remember:
Analizar\Comparar medias\ANOVA de un
factor
Small p-value:
We reject Ho. Anyway, whether Ho is rejected, we can’t considere
that all means are different. So rejecting Ho means that at least two
means have different values. Then we need tools to perform a
multiple comparison procedure (Post Hoc), such as:
We can conclude that
Channels 1 and 2 are considered equals.
Now, we are
available to present our conclusions as an
“Image”, such results are a complement that can be easy
incorporate to a Dashboard within Oracle Business Intelligence,
using: fmap:Images/[Image.Name].[Image.Extension] (selecting "Link or Image" in the Dashboards Objects).
Another posibility
(in order to display an Image) is using
of calculated column within a metric performed with Oracle Answers, such as:
'img src="Images/'|| [Column with Image.Name and Image.extension] || '"/'
And as a result:
Hope it helps!
Best wishes.
Joan Martí Peraire.
