Data Mining Training gives No Results

[ist]

Hi,
I am studying data mining features of SSAS and for a workshop I've
created 2 views derived from vTargetMail view of AdventureWorksDW.
Train data consists every record except those in Pacific, and test
view consists only records from Pacific area.

1. I've created a mining structure based on N.Bayes alg. and selected
BikeBuyer as predictable column.
2. According to input column suggestions, I've selected Age,
Eng.Education, NumberCarsOwned, YearlyIncome, CommuteDistance,
NumberChildsatHome and TotalChildren as input columns
3. I've modified no other setting, and deployed project.


However all I get is a dependency network with a single predict
column, empty attribute prof./char./disc. tables, an empty
classification matrix and a lift chart with only 45 degree line.


Am I missing a step, or must I do some fine-tuning on (what)
parameters?


Thanks...

[Dejan Sarka]

Hi!

I checked and I got the same result. It seems that the wizard does not
detect automatically content type of attributes. During the wizard, there is
a button called "Detect" in one of the pages; if you click on it, it detects
the content type better (still not perfect). In my case (without clicking on
Detect), content of all attributes (the Content property) was Discretized,
with empty DiscretizationBucketCount and DiscretizationMethod properties;
these settings make no sense. Therefore, it is logical the algorithm does
not find any interesting dependency.
You can change the Content property of attributes. You can do it in the
Designer window, using the Mining Structure tab (the first tab). Click on
each column and check the Content, DiscretizationBucketCount and
DiscretizationMethod properties. You can set these properties like this:

Column, Content, DiscretizationBucketCount, DiscretizationMethod
Age, Discretized, 5, Automatic
Bike Buyer, Discrete, blank, blank
Commute Distance, Discrete, blank, blank
English Education, Discrete, blank, blank
Number Cars Owned, Discrete, blank, blank
Number Children At Home, Discrete, blank, blank
Total Children, Discrete, blank, blank
Yearly Income, Discretized, 5, Automatic

Also, do not forget you need a key. You should use Customer Key as the key
(Content should be Key).

Change these properties and you should get some meaningful patterns.
--
Dejan Sarka
http://blogs.solidq.com/EN/dsarka/

Post by ist
Hi,
I am studying data mining features of SSAS and for a workshop I've
created 2 views derived from vTargetMail view of AdventureWorksDW.
Train data consists every record except those in Pacific, and test
view consists only records from Pacific area.
1. I've created a mining structure based on N.Bayes alg. and selected
BikeBuyer as predictable column.
2. According to input column suggestions, I've selected Age,
Eng.Education, NumberCarsOwned, YearlyIncome, CommuteDistance,
NumberChildsatHome and TotalChildren as input columns
3. I've modified no other setting, and deployed project.
However all I get is a dependency network with a single predict
column, empty attribute prof./char./disc. tables, an empty
classification matrix and a lift chart with only 45 degree line.
Am I missing a step, or must I do some fine-tuning on (what)
parameters?
Thanks...

[ist]

Hi,

Thanks for your interest, in fact I had deleted this post and re-
posted to other groups with some change. Sorry if there seemed some
duplicate posts.

Ok, I have tried applying your suggestions. I've discretized some
numeric columns and reprocessed the model. However, I still get the
same result.

If possible; can you please send your project files to me? So that I
can compare it with my project and find my missed steps/configurations
(saygin_at_gmail___com)

Thanks again...

On 5 Eyl�l, 18:43, "Dejan Sarka"

Post by Dejan Sarka
Hi!
I checked and I got the same result. It seems that the wizard does not
detect automatically content type of attributes. During the wizard, there is
a button called "Detect" in one of the pages; if you click on it, it detects
the content type better (still not perfect). In my case (without clicking on
Detect), content of all attributes (the Content property) was Discretized,
with empty DiscretizationBucketCount and DiscretizationMethod properties;
these settings make no sense. Therefore, it is logical the algorithm does
not find any interesting dependency.
You can change the Content property of attributes. You can do it in the
Designer window, using the Mining Structure tab (the first tab). Click on
each column and check the Content, DiscretizationBucketCount and
Column, Content, DiscretizationBucketCount, DiscretizationMethod
Age, Discretized, 5, Automatic
Bike Buyer, Discrete, blank, blank
Commute Distance, Discrete, blank, blank
English Education, Discrete, blank, blank
Number Cars Owned, Discrete, blank, blank
Number Children At Home, Discrete, blank, blank
Total Children, Discrete, blank, blank
Yearly Income, Discretized, 5, Automatic
Also, do not forget you need a key. You should use Customer Key as the key
(Content should be Key).
Change these properties and you should get some meaningful patterns.
--
Dejan Sarkahttp://blogs.solidq.com/EN/dsarka/

Post by ist
Hi,
I am studying data mining features of SSAS and for a workshop I've
created 2 views derived from vTargetMail view of AdventureWorksDW.
Train data consists every record except those in Pacific, and test
view consists only records from Pacific area.
1. I've created a mining structure based on N.Bayes alg. and selected
BikeBuyer as predictable column.
2. According to input column suggestions, I've selected Age,
Eng.Education, NumberCarsOwned, YearlyIncome, CommuteDistance,
NumberChildsatHome and TotalChildren as input columns
3. I've modified no other setting, and deployed project.
However all I get is a dependency network with a single predict
column, empty attribute prof./char./disc. tables, an empty
classification matrix and a lift chart with only 45 degree line.
Am I missing a step, or must I do some fine-tuning on (what)
parameters?
Thanks...- Al nt y gizle -

- Al nt y g�ster -

[Dejan Sarka]

Sent.
--
Dejan Sarka
http://blogs.solidq.com/EN/dsarka/

"ist" <***@gmail.com> wrote in message news:***@o80g2000hse.googlegroups.com...
Hi,

Thanks for your interest, in fact I had deleted this post and re-
posted to other groups with some change. Sorry if there seemed some
duplicate posts.

Ok, I have tried applying your suggestions. I've discretized some
numeric columns and reprocessed the model. However, I still get the
same result.

If possible; can you please send your project files to me? So that I
can compare it with my project and find my missed steps/configurations
(saygin_at_gmail___com)

Thanks again...

On 5 Eylül, 18:43, "Dejan Sarka"

Post by Dejan Sarka
Hi!
I checked and I got the same result. It seems that the wizard does not
detect automatically content type of attributes. During the wizard, there is
a button called "Detect" in one of the pages; if you click on it, it detects
the content type better (still not perfect). In my case (without clicking on
Detect), content of all attributes (the Content property) was Discretized,
with empty DiscretizationBucketCount and DiscretizationMethod properties;
these settings make no sense. Therefore, it is logical the algorithm does
not find any interesting dependency.
You can change the Content property of attributes. You can do it in the
Designer window, using the Mining Structure tab (the first tab). Click on
each column and check the Content, DiscretizationBucketCount and
Column, Content, DiscretizationBucketCount, DiscretizationMethod
Age, Discretized, 5, Automatic
Bike Buyer, Discrete, blank, blank
Commute Distance, Discrete, blank, blank
English Education, Discrete, blank, blank
Number Cars Owned, Discrete, blank, blank
Number Children At Home, Discrete, blank, blank
Total Children, Discrete, blank, blank
Yearly Income, Discretized, 5, Automatic
Also, do not forget you need a key. You should use Customer Key as the key
(Content should be Key).
Change these properties and you should get some meaningful patterns.
--
Dejan Sarkahttp://blogs.solidq.com/EN/dsarka/

Post by ist
Hi,
I am studying data mining features of SSAS and for a workshop I've
created 2 views derived from vTargetMail view of AdventureWorksDW.
Train data consists every record except those in Pacific, and test
view consists only records from Pacific area.
1. I've created a mining structure based on N.Bayes alg. and selected
BikeBuyer as predictable column.
2. According to input column suggestions, I've selected Age,
Eng.Education, NumberCarsOwned, YearlyIncome, CommuteDistance,
NumberChildsatHome and TotalChildren as input columns
3. I've modified no other setting, and deployed project.
However all I get is a dependency network with a single predict
column, empty attribute prof./char./disc. tables, an empty
classification matrix and a lift chart with only 45 degree line.
Am I missing a step, or must I do some fine-tuning on (what)
parameters?
Thanks...- Al nt y gizle -

- Al nt y göster -

[ist]

Received & replied :)
Thank you very much Mr.Sarka.

For anyone who is interested; the problem is solved by changing my
computer's regional and non-unicode programs configurations to US
English (which were set to Turkish).

On 6 Eyl�l, 09:04, "Dejan Sarka"

Post by Dejan Sarka
Sent.
--
Dejan Sarkahttp://blogs.solidq.com/EN/dsarka/
Hi,
Thanks for your interest, in fact I had deleted this post and re-
posted to other groups with some change. Sorry if there seemed some
duplicate posts.
Ok, I have tried applying your suggestions. I've discretized some
numeric columns and reprocessed the model. However, I still get the
same result.
If possible; can you please send your project files to me? So that I
can compare it with my project and find my missed steps/configurations
(saygin_at_gmail___com)
Thanks again...
On 5 Eyl�l, 18:43, "Dejan Sarka"

Post by Dejan Sarka
Hi!
I checked and I got the same result. It seems that the wizard does not
detect automatically content type of attributes. During the wizard, there is
a button called "Detect" in one of the pages; if you click on it, it detects
the content type better (still not perfect). In my case (without clicking on
Detect), content of all attributes (the Content property) was Discretized,
with empty DiscretizationBucketCount and DiscretizationMethod properties;
these settings make no sense. Therefore, it is logical the algorithm does
not find any interesting dependency.
You can change the Content property of attributes. You can do it in the
Designer window, using the Mining Structure tab (the first tab). Click on
each column and check the Content, DiscretizationBucketCount and
Column, Content, DiscretizationBucketCount, DiscretizationMethod
Age, Discretized, 5, Automatic
Bike Buyer, Discrete, blank, blank
Commute Distance, Discrete, blank, blank
English Education, Discrete, blank, blank
Number Cars Owned, Discrete, blank, blank
Number Children At Home, Discrete, blank, blank
Total Children, Discrete, blank, blank
Yearly Income, Discretized, 5, Automatic
Also, do not forget you need a key. You should use Customer Key as the key
(Content should be Key).
Change these properties and you should get some meaningful patterns.
--
Dejan Sarkahttp://blogs.solidq.com/EN/dsarka/

Post by ist
Hi,
I am studying data mining features of SSAS and for a workshop I've
created 2 views derived from vTargetMail view of AdventureWorksDW.
Train data consists every record except those in Pacific, and test
view consists only records from Pacific area.
1. I've created a mining structure based on N.Bayes alg. and selected
BikeBuyer as predictable column.
2. According to input column suggestions, I've selected Age,
Eng.Education, NumberCarsOwned, YearlyIncome, CommuteDistance,
NumberChildsatHome and TotalChildren as input columns
3. I've modified no other setting, and deployed project.
However all I get is a dependency network with a single predict
column, empty attribute prof./char./disc. tables, an empty
classification matrix and a lift chart with only 45 degree line.
Am I missing a step, or must I do some fine-tuning on (what)
parameters?
Thanks...- Al nt y gizle -

- Al nt y g�ster -- Al nt y gizle -

- Al nt y g�ster -