Question on Value of Probability of value 1 or 2 in Neural Network Viewer

Hi, all,

I am confused about the value of Probability of Value 1 or 2 (on a particular attribute value) in Neural Network viewer. E.g. the value of Probability of value 1 is actually very low (the same to the value of Probability of value 2), but why the bar which shows the strength of the probability of these two values are still so strong even stronger than other values of probability of value 1 or 2 based on other attribute values which have a much higher probability of value 1 or 2?

And how does the algorithm calculate the Probability of attribute value in nerual network by the way?

Hope my question is clear.

I am looking forward to hearing from you shortly and thanks a lot in advance.

With best regards,

Yours sincerely,

The scores represent the predicted probability as influenced by a single attribute value.

Here is how the scores are computed:

Assume that the model predicts Bike Buyer (= TRUE or FALSE), based on Home Owner, Commute Distance etc.

A stored procedure creates a "virtual" input, containing, for each row, a single attribute (everything else is missing). Example:

Row 1: (empty) -- all attributes are missing

Row 2: Home Owner = TRUE, everything else is empty (missing)

Row 3: Home Owner = FALSE, everything else is empty (missing)

Row 4: Commute Distance= 0-2 miles, everything else is empty (missing)

Row 5: Commute Distance= 2-4 miles, everything else is empty (missing)

and so on.

For each row in this "virtual" input, the procedure computes PredictProbability(BikeBuyer, 'TRUE') and PredictProbability(BikeBuyer, 'FALSE') as well as Predict(BikeBuyer)

All the results that favor TRUE are then normalized in the 0-100 space

Same for all the results that favor FALSE

So, practically, the Probability of Value 1(or 2) represents a normalized view of the prediction probability for value 1 generated by that input.

Because of the normalization, inputs that predict strongly a value may still have low scores (simply because all other inputs are stronger)

Hope this helps

|||

Hi,

Thanks for that.

With best regards,

sql

Question on filter itemset by up to one attribute in Association Rules

Hi, all experts here,

Thanks for your kind attention.

I want to filter the itemsets or rules based on more than 2 attributes, how can we achieve that? (I can only filter them by only one attribute?). Is it possible to achieve that?

Thanks a lot and I am looking forward to hearing from you shortly.

With best regards,

Yours sincerely,

The filter actually can specify a regular expression, using the .Net language for regular expressions (the RegEx class).

Assume that you want any items set that include the Helmets and Fenders attributes, in any order

Here is an example of such filter:

((.*Helmets.*Fenders.*)|(.*Fenders.*Helmets.*))|||

Hi, are you saying this can be achieved in the association rules viewer?

Thanks.

|||Yes. In the filter, type directly the regular expression|||

Hi, Bogdan,

Thanks a lot.

Best regards,

Question on attribute with more than one column

Hi,all experts here,

Thank you very much for your kind attention.

I am having a question on attribute with more than one column in a cube. What is the benefitial to have a attribute with more than one column? I am confused for this conception. And could please any experts here shed me any light on that.

Thanks a lot in advance for your help and I am looking forward to hearing from you.

With best regards,

Yours sincerely,

I'm not sure if this is the question that you are asking, but, here's a stab at the answer...

Often multiple columns in a relational table can be used to uniquely identify each member in an attribute. For example - months are repeated in years. If you have a Month column in your table that lists all the months by their names - Jan, Feb and so on - you won't be able to distinguish between months belonging to year 2003 and months belonging to year 2004. Now if you have another column (say Year) that lists the year, then, you could have a single attribute that has a composite key made up of the columns Year and Month.

|||

Hi, Bala,

Thanks a lot.

With best regards,

Yours sincerely,

Question on attribute with more than one column

Hi,all experts here,

Thank you very much for your kind attention.

I am having a question on attribute with more than one column in a cube. What is the benefitial to have a attribute with more than one column? I am confused for this conception. And could please any experts here shed me any light on that.

Thanks a lot in advance for your help and I am looking forward to hearing from you.

With best regards,

Yours sincerely,

I'm not sure if this is the question that you are asking, but, here's a stab at the answer...

Often multiple columns in a relational table can be used to uniquely identify each member in an attribute. For example - months are repeated in years. If you have a Month column in your table that lists all the months by their names - Jan, Feb and so on - you won't be able to distinguish between months belonging to year 2003 and months belonging to year 2004. Now if you have another column (say Year) that lists the year, then, you could have a single attribute that has a composite key made up of the columns Year and Month.

|||

Hi, Bala,

Thanks a lot.

With best regards,

Yours sincerely,

Question on attribute relationships and hierarchies on SQL Server 2005 AS Cubes

Hi, all experts here,

Thank you very much for your kind attention.

I am having a question on the attribute relationships and hierarchies in the cubes. It is a must to create the attribute relationships for each levels in any hierarchies in order to let the UDM to get the correct aggregates over different hierarchies? (e.g. within hierarchy H, levels are a-b-c-d. Do we have to create the attribute relationships for all the levels?, e.g. we have to create the attribute relationship between a and b, b and c, c and d? ). Hope my question is clear for your help and I am looking forward to hearing from you shortly.

Thanks a lot.

With best regards,

Yours sincerely,

Hello. Correct! Your cubes will benefit in performance if you add attribute relations. The way you have described this is correct but you will only need to create the in your natural or user hierarchies(the middle pane in the dimension editor).

HTH

Thomas Ivarsson

|||

Hi, Thomas,

Thanks a lot.

With best regards,

Yours sincerely,

Question on attribute keys of measure group

Hi, dear friends,

I encountered a problem processing my cube with the error message telling: processing measure_group_name (which only has a count as its only measure)failed as attribute keys can not be found? Why is that? I have no idea why is that? Does it mean I have to include all the keys of the fact table where the measure group from in the measure groups as measures? Otherwise the system failed to identify the distinct count of the system? Or whatever measures the measure group contains, it alwys have to include all the keys including the dimension keys of dimensions which the measure group is related to?

Thanks in advance for your help and advices and I am always looking forward to hearing from you shortly from you.

With best regards,

Yours sincerely,

This message means that a column used to join the measure group to a dimension contains a value not found in the dimension. This is usually caused by a referential integrity error - the FK does not existing in the PK of the lookup table. However, it can also be caused by the presence of a null in the FK as by default this null will be converted to zero or blank which probably does not exist as a key in the dimension. You can handle the situation of nulls in FK columns within the fact table by setting the NullProcessing option in the advanced tab of the measure group relationship dialog from the Dimension Usage tab of the cube editor.|||

Hi, Matt,

Thanks a lot. Yes, the problem is the matching of the PK and the FK between the tables.

However, the data source view logical key and the relationships between tables are really annoying then, as it is not able to check the integrity of the tables?

With best regards,

Yours sincerely,

|||

So it sounds like you would like to have an option on the Data Source View to validate logical keys and relationships. I'd recommend you file a feature request on http://connect.microsoft.com/SQLServer/Feedback. Such feature requests will be automatically added to our issue tracking system for consideration in future releases and those that originate from a customer such as yourself get extra consideration since we know it's a real world scenario. If you can describe the scenario it will also help as it will allow us to better identify which requests will help the most people and how to implement them so that they actually provide the help intended.

Thanks, Matt

|||

Hi, Matt,

Thanks a lot and this sounds really a good idea of it.

With best regards,

Yours sincerely,