Dangers in Dashboard Design

Dangerous Dashboard Design

Dangerous Dashboard Design

I recently had the great pleasure of working on a dashboard project using PerformancePoint in SharePoint 2013 with a great team of individuals.  We had to overcome several obstacles during the process and I thought it would be a good idea to document how we accomplished some of our goals with the hope that I can spare someone else (or my future self) from having to go through the same pain.  Our dashboard is currently in closed beta and we will be releasing it to the wild in the next couple of weeks.

Dynamic Active Directory Security Without Kerberos 

The biggest challenge we faced was how to implement dynamic active directory security without using Kerberos.  I had applied dynamic security to several Analysis Services databases in the past; but, had never followed that implementation through to PerformancePoint dashboards.  This challenge is more difficult in that respect because within a dashboard you will typically have more sources of data than just an analysis services database.  We had SSRS reports that ran against our ODS and our relational data warehouse.  You can read more about how we achieved this in my recent blog post, PerformancePoint 2013 Dashboards: Applying Dynamic Security using CustomData and EffectiveUserName.

Our solution required three roles.  Two roles for the managers who would be restricted to seeing only their employees.  One role for the PerformancePoint objects which uses the CustomData function in the security and another for the SSRS MDX based reports and Excel Services based reports that uses the UserName function.   Finally we needed a separate unrestricted role for the managers and district leaders that could see all of the data.  This allowed us to keep our bridge table relatively small.  Otherwise every person that needed to see every person would require about 2,500 rows in the bridge table.

During develoment, I mistakenly thought we could create the entire dashboard for the restricted users, save a copy and rename it to unrestricted, change the data source to use the unrestricted role and be on our way.  I was woefully mistaken.  Apparently PerformancePoint uses some sort of invisible GUID or something for the data source name.  Worse yet, you can’t change the data source for existing Analytic charts, grids, and scorecards.  So we couldn’t make a copy of the objects, rename them, and point them to a different data source.  We actually had to use two different site collections for the restricted and the unrestricted dashboards!

I Know Who You Are and What You Did Last Quarter

Once we had the dynamic security in place we wanted the dashboard to automatically know who you are.  We accomplished this by adding an attribute to our person dimension that would hold the active directory account name.  I use the word person in it’s most generic sense.  A person can be a vendor, customer, employee, manager, or sales person.  I created a MDX filter in PerformancePoint and used the following MDX code to achieve this: 

STRTOSET(“[People].[Active Directory Login].&[” + CUSTOMDATA() + “]”)

I then added this attribute to any analytic charts, grids, or scorecards in the background quadrant of the design canvas so that the report could be filtered by it.  I connected the filter to the report as one typically would and later hid the filter using the edit web part interface within the SharePoint 2013 site.

Save the Trees

My good buddy Michael Simon used some fancy MDX to set the trailing six quarters for an analytic bar chart only to discover that we lost the ability to right click and get the decomposition tree and all the other neat PerformancePoint features.  He discovered a clever work around using dynamic named sets.  We had our right clicks back.

Yesterday… Not Such an Easy Game to Play

Another challenge we faced is a very common one.  How to get the dashboard to automatically present data as of yesterday.  Typically this wouldn’t be a big deal; but, in our case we use a reporting calendar that doesn’t include Saturdays, Sundays or Holidays.  Any sales that occur on those days roll forward to the next business day.  If you don’t have such a dilemma you can use either of these two snippets to accomplish this feat:

1. STRTOSET(“[Date].[Calendar].[Date].&[“+ VBAMDX.Format(VBAMDX.Now()-1, “yyyyMMdd”) +”]”)

2. STRTOSET(“[Date].[Calendar].[Date].&[“+ VBAMDX.Format(VBAMDX.Now(), “yyyyMMdd”) +”].Lag(1)”)

In our case, this wouldn’t work.  If the user wanted to look at the dashboard on a weekend or a holiday they would get an error. 

I initially tried to set the default member of the dimension to yesterday using MDX.  I ran into a problem in that once I set the default member for the date dimension in the cube I could no longer put the date dimension in the report filter area of an Excel Pivot table.  It would no longer work and would just show yesterday’s data.  See this link for more information about that issue:  Default Member Trap (be careful of what you filter)

My next strategy was to add an attribute to our Gregorian date dimension called Last Business Date and update the corresponding relational table with this information.  I then attempted to parse together the information to get the correct reporting date.  We have two date dimensions in our cube.  One is a regular calendar date dimension, the other one is a fiscal reporting date dimension that doesn’t include weekends or holidays.  Our fact table has keys for both.  You can see my post asking for help from the BIDN community here: 

MDX Help Please – The + function expects a tuple set expression for the 1 argument. A string or numeric expression was used

Before I received the kind reply from mreedsqlbi, I found another way to accomplish what I needed.  This post helped me find another solution:  Member Does Not Exist in MDX Calculations

I created four dynamic named sets in the cube called, OneDayAgo, TwoDaysAgo, ThreeDaysAgo, and FourDaysAgo.

Here is the code for OneDayAgo:

iif(iserror(STRTOMEMBER(“[Date – Reporting].[Date].&[” + VBAMDX.Format(VBAMDX.Now()-1, “yyyyMMdd”) +”]”)),{},STRTOMEMBER(“[Date – Reporting].[Date].&[” + VBAMDX.Format(VBAMDX.Now()-1, “yyyyMMdd”) +”]”))

For each of the other named sets, I just changed the –1 to a –2, –3, and –4 respectively.

The code for my MDX filter used the dynamic named sets to figure out if the member exists:


IIF([OneDayAgo].Count > 0
            ,STRTOMEMBER("[Date - Reporting].[Date].&[" + VBAMDX.Format(VBAMDX.Now()-1, "yyyyMMdd") +"]")
            ,IIF([TwoDaysAgo].Count > 0
                  ,STRTOMEMBER("[Date - Reporting].[Date].&[" + VBAMDX.Format(VBAMDX.Now()-2, "yyyyMMdd") +"]")
                  ,IIF([ThreeDaysAgo].Count > 0
                        ,STRTOMEMBER("[Date - Reporting].[Date].&[" + VBAMDX.Format(VBAMDX.Now()-3, "yyyyMMdd") +"]")
                        ,IIF([FourDaysAgo].Count > 0
                            ,STRTOMEMBER("[Date - Reporting].[Date].&[" + VBAMDX.Format(VBAMDX.Now()-4, "yyyyMMdd") +"]")
                            ,STRTOMEMBER("[Date - Reporting].[Date].&[" + VBAMDX.Format(VBAMDX.Now()-5, "yyyyMMdd") +"]")
                        )
                    )
                )
            )

UPDATE:

This works better because it works in Dev which doesn’t always have data for yesterday.

TopCount(ORDER( NONEMPTY( {[Date – Reporting].[Date].Members

     – [Date – Reporting].[Date].[All]}

   , [Measures].[Sales Amt] )

   , [Date – Reporting].[Date].CurrentMember.MEMBER_KEY

   , DESC )

,1)

UPDATE:

At one point during Beta testing, I was asked to present the user with a date dropdown that defaults to yesterday.  This was later scrapped because it required a server wide change within the Central Administration Application Settings for PerformancePoint in which we had to set the user selected filters to expire every day.

For anyone who is interested, I got that to work by removing the TOPCOUNT from the MDX formula filter in Dashboard Designer:

ORDER( NONEMPTY( {[Date – Reporting].[Date].Members – [Date – Reporting].[Date].[All]},                 
 [Measures].[Sales Amt] ), [Date – Reporting].[Date].CurrentMember.MEMBER_KEY, DESC )

This only shows dates that have data, excludes the all member, and sorts it in descending order.  The result is that it defaults to yesterday but allows you to change it. 

UPDATE

I think I’m crazy; but, I could have sworn the dashboard was showing Friday’s data on Monday.  Looking at the code, I don’t see how; but, I am amazed no one has pointed it out.

I had to modify the code to eliminate Today’s date from the data set.

TopCount
  (
    Order
    (
      NonEmpty
      (
        {
          [Date – Reporting].[Date].MEMBERS
   – [Date – Reporting].[Date].[All]
   – STRTOMEMBER(“[Date – Reporting].[Date].&[” + VBAMDX.Format(VBAMDX.Now(), “yyyyMMdd”) +”]”)
        }
       ,[Measures].[Sales Amt]
      )
     ,[Date – Reporting].[Date].CurrentMember.Member_Key
     ,DESC
    )
   ,1
  )

Filters

Another challenge we faced was how to have MDX filters control relational reports.  This was pretty easy.  Just use the DisplayName attribute of the filter instead of MemberUnique and let the report figure out how to use that information.  We were unable to use multi-select MDX member selection filters.  When we passed these as DisplayValue to our relational SSRS reports it would only pass the first selection.  We were unable to overcome this limitation the way we would have liked.  We ended up not using multi-select filters and instead had filters that only contained the parent level of the hierarchy.  In other words, our filter contained managers and the report would display the managers employees.  We lost the ability to have the user only select some employees of that manager; but, not others.  To my knowledge, there is not a way to pass MDX multi-value filters using the DisplayValue property to relational SSRS reports.

Staying On Top of Reporting Services

One aggravation we experienced was that the column headers on the SSRS reports would not stay at the top of the web part when the user scrolled through the report part.  There are many blog posts out there on how to achieve this; but, within SharePoint 2013 it would work in Firefox and Chrome; but, not in Internet Exploder (9 or 10.)  I had to workaround this issue by sizing the report so it wouldn’t scroll, and instead display the toolbar and have the user page through the data instead of scroll.  Which in my opinion, looked nicer anyway.

I’ve got to mention one gotcha we experienced during our dashboard design endeavor that I had experienced many years ago.  I thought I had blogged about it in my What’s New BI-wise section; but, I couldn’t find it so I must have dreamed I had blogged about it.  Any way here are some words of wisdom:

When you are creating an SSRS MDX based report and need to make a customization to the underlying MDX that is created using the easy to use drag and drop interface, make a backup or better yet, check it into source control and label it ‘Pre-MDX customizations’ or something very clear like that.

Once you have made those customizations to the MDX (to allow that member property to be displayed for example) you can’t go back!  No longer can you simply delete all your parameters and add different ones easily.  Everything has to be done through hand coded MDX from that point forward.  It is far easier to design your report as much as possible, make a backup or labeled source control check-in, and then make your MDX customizations.  If you need to add/remove fields or parameters in the future it is easier to go back to that pre-MDX customized version and do it using the interface and then re-do the MDX customization than to work with it outside the interface.

Scorecards Don’t Always Score 

Scorecards are really cool gadgets to use in PerformancePoint dashboards because you can use them to control other objects.  Filters and scorecards are the only way to control objects in the PerformancePoint dashboard.  When you see a scorecard functioning well inside a PerformancePoint dashboard it looks really cool.  As you click on the various members within the scorecard you can have charts and data change according to the context.

We could not use scorecards the way we wanted because we simply had too many members in our dimension for them to render quickly enough.  They don’t work as you would expect.  Doing the same type of thing in an analytic grid or Excel services report is a piece of cake and renders very fast.  For whatever reason, regardless of the fact that I filtered the scorecard on the manager, it would insist on rendering every employee in the company and then hiding the ones that weren’t children of the selected manager.  In our case that was a little less than 5,000 members.  I reduced this set to about 2,800 by deleting employees that weren’t used as keys in any of our fact tables; but, it was too late.  The powers that be had already decided that scorecards stink due to not only the rendering speed; but, the hoops we had to jump through to develop one using that many members.  Even in the developer’s IDE, Dashboard Designer, it would pull back every single representative regardless of being filtered on a single parent member.  It would display those in a red font indicating that they would be hidden; but, the time it took the designer to render made you think it had crashed.  We actually had to build a cube containing a very small subset of the data in order to design the scorecard the way we wanted.

From a pure developer’s standpoint it seems that the scorecard design process in PerformancePoint needs some maturing.  Some configurations that are set within the wizard for creating an Analysis Services based scorecard are simply not editable outside of the wizard.  One would need to take screenshots of how the scorecard was developed to document the process in the event that a change was ever needed to one of those initial configurations.

While I’m on the subject of scorecards, another gripe we had about the way scorecards work is that any time you have more than one set of KPI actuals and targets within a scorecard it displays them with all of the actuals on the left and trends or targets on the right.  In other words, if I have a sales actual and goal KPI on the same scorecard as a gross profit actual and goal, I can only show them as actual sales, actual gross proft, target sales, target gross profit.  I would like some control to show the targets next to their corresponding actuals. 

Easy Migrations 

One new feature of SharePoint 2013 that I really like is the new way it handles exporting and importing a dashboard project.  In previous versions of SharePoint, I had to create long lists of customizations that had to be done to the web parts after Dashboard Designer had deployed the dashboard.  Things such as hiding filters, resizing objects, and changing the chrome settings had to be done over and over again every time we made a change to the dashboard that required a re-deployment.  Moving the project from the development environment to the QA environment and finally to the production environment was a manual tedious process.  Furthermore, these customizations were not stored in source control.  I was pleased as punch to discover the export/import feature of SharePoint 2013 solves both problems.  I can extract the resulting .CMP file using a cab extractor and put the entire project into source control and I don’t have to repeat those customizations over and over again.

Accepting the Unacceptable 

I’ve been creating dashboards since PerformancePoint 2007.  Make no mistake, I do not consider myself an expert at dashboard design.  I would however consider myself pretty advanced in using the Microsoft BI stack to produce great dashboards quickly.  I really like using PerformancePoint.  If you are willing to accept some of the quirks w/ PerformancePoint, it is a great tool to allow a developer to very quickly produce meaningful dashboards with a lot of functionality right out of the box.  When you start getting very particular about the way things look it gets more complicated.  For example, I can only show the legend at the top or to the right of my analytic charts.  I can’t show them on the left and I can’t show them on the bottom.  I can’t control the color of the bars in the bar charts.  I sometimes joke that designing PerformancePoint dashboards is a great opportunity for someone who likes to say no.  Sure, we can give you a bar chart w/ the colors you want or with the legend on the left; but, I have to do that in Reporting Services and we lose all of the cool right click functionality that is included in the analytic chart such as the decomposition tree. 

 

Fifty Shades Of Gray Is Not Enough 

I’ll never forget one project I worked on as a consultant with a very lovely lady who had designed her dashboard on paper before she had any idea of what PerformancePoint could and could not do easily.  A simple thing like showing the total dollar amount at the top of the bar chart involved creating a separate reporting services web part that only showed that total and placing it above the chart. The end result was a really good looking dashboard that looked remarkably similar to what had been envisioned.  It just took a lot of extra time to make it look exactly how it was imagined.

One other piece of advice I can give is to apply a lot of rigor in your testing.  Make sure each page of the dashboard uses consistent nomenclature.  Make sure that the numbers on all pages agree with each other.  If one page says that gross profit is this and another says gross profit was that, you’re going to get a lot of questions from your user community.  Be consistent in your color and number formatting as well.  Are you going to display zeros or leave the cells blank?  I personally prefer to leave the cells blank; but, it is more important that whatever you decide you stay consistent.  Decide beforehand what screen resolution you are designing to.  If everyone at your company has widescreen 17 inch monitors than it might be best to design your dashboard to look best at 1920×1080.  If you need to show your dashboard on an iPad or a Surface than you better have one lying around.  Good luck with that by the way.  Consider which browser is your company’s standard.  HTML5 or not, things look different in Chrome, Firefox, Safari and Internet Exploder versions 8, 9, 10, or 11.  

Awwww…

As I wrap this up, I just want to give a big warm thank you to the other members of my team, Sherri McDonald and Michael Simon.  A lot of hard work and long hours went into this project and it feels great to have professionals like them to ease the burden.  It’s a much better experience than working solo.  Three heads are much better than one and it feels great to have others throw rocks at your stuff and see where it breaks before the real critics get their hands on it.

Do you have a good dashboard design story?  Did I leave a danger out?  Please let us know in the comments. 

——————————————————————————- 

Other Dashboarding Resources

One person that I would consider an expert on dashboard design is Stephen Few.  I recently read on his blog that a second edition of his book, Information Dashboard Design: The Effective Visual Communication of Data, is being released shortly.  He mentioned in that post that one of the criticisms he received on the first edition that is addressed in the second edition is the lack of real examples of effective dashboards by real products.  His explanation is that at that time there simply weren’t any products released that were capable of producing what he considered really good dashboard designs.  Be sure to read his article Common Pitfalls in Dashboard Design for a great list of dashboard design gotchas.

Links

PerformancePoint 2013 Dashboards: Applying Dynamic Security using CustomData and EffectiveUserName

What are Web Parts?

PerformancePoint – Effectively Disable Analytic Grid Right Click Select Measures Functionality

Time Intelligence Filters in PerformancePoint 2010

Cascading Filters in Performance Point Services dashboard using SharePoint 2013

Performance Point Relative Date Time Intelligence with current date time

How to make SSRS reports in a PerformancePoint dashboard ‘pop out’ or open in a new window.

Clutter, data overload put dashboard designs on path to failure

Data Visualization and Dashboard Design

Formatting your PerformancePoint Analytic Grid!

Using SSAS MDX Calculation Color Expressions

Time Intelligence Post Formula Filters

Set Default Value of PerformancePoint Filter – Note:  Caveat to this tip- Once a user changes the default value SharePoint will remember the changed value and not reset the default value.

Add conditional formatting to your PerformancePoint Services Analytic Grid, by defining a custom Scope calculation from within Analysis Services (SSAS)

Filter Connection Quick Reference

——————————————————————————-

All my PerformancePoint links – Updated as I update them!

——————————————————————————-

PerformancePoint Gotcha!  – “PerformancePoint Services could not connect to the specified data source.”  When creating a new dashboard, you MUST do a save all right after you create your data sources otherwise you won’t be able to create anything that uses that data source.

——————————————————————————-

Gotcha! PerformancePoint Time Intelligence and SSRS

I was trying to use a PerformancePoint time intelligence filter to pass values to a SSRS report. I discovered that the MonthToDate, YearToDate, QuarterToDate syntax does not work with SSRS. Instead use Month.FirstDay:Day, Year:FirstDay:Day, Quarter.FirstDay:Day

——————————————————————————-

Zero to Dashboard- Intro to PerformancePoint    
Come check out the PerformancePoint hotness! Mike will demonstrate the functionality in PerformancePoint services in SharePoint 2010 and show you how to quickly build some dynamic dashboards and reporting functionality your end users will crave.

http://pragmaticworks.com/Resources/webinars/WebinarSummary.aspx?ResourceID=327

——————————————————————————-

PerformancePoint Tip – Right clicking an analytic grid in the leftmost columns allows the user to select measures.  Some people may not want that, so how do you disable that ability?  PerformancePoint Services Application Settings – Select Measures Control – set maximum value to 0.  (default is 1000).

http://technet.microsoft.com/en-us/library/ee620542.aspx

——————————————————————————-

Advertisements

Data Mining with DMX

To my knowledge, Microsoft has three methods of performing data mining within their BI stack.  The Excel Add-In, the GUI and wizards within BIDS, and the DMX query language.  Obviously the key draw back to the Excel Add-in is that you can’t schedule it to run automatically without human intervention.  However, the Excel Add-in provides an excellent method to form your hypothesis and test it.  The GUI and the wizards are nice and you can create repeatable processes using those tools alone.  The command line interface that DMX provides has it’s own appeal to old folks like me who can remember when computer magazines included program listings.

There are basically four steps to data mining with DMX:

  1. Creation of Structures
  2. Creation of Mining Models
  3. Training
  4. Prediction

Think of structures as you would a framework for a house.  Now imagine there are those half mannequins scattered throughout the house structure.  The ones without the arms, legs, and head.  Just the torso in a skirt.  These mannequins are clothed with cheap garments.  The physical shape of the models is determined by the various data mining algorithms.  Some are fit, some are chunky, some are pencil thin.  Next imagine tons of paint being dumped on those garments by the bucket.  One color for each bucket.  At the end of the day, the garments are going to contain patterns of colors streaking down them derived from the buckets of different colored paints.  Those are your models.  Now, imagine a sophisticated computer is attached to those models.  Based on the colors of paint that were dumped on the garments during their training, we can predict the streaks of colors that will occur if an entirely new batch of paint is dumped upon those models.  That is prediction.  This is DMX.  DMX provides the language that defines the structures, models, and querying language.  Now imagine there is a purple cat in the corner of the house drinking lemonade.

The best thing about DMX is that it so closely resembles SQL.  They tried to do that with MDX; but, conceptually it just doesn’t fit quite right.  Knowing SQL can be a hindrance to knowing MDX.  Not so with DMX.

The most commonly used data types in DMX closely resemble those that we encounter with SQL:  Long, double, text, date, Boolean, and table.  Types in SQL that are not in DMX, such as integer, money, float, etc. are implicitly converted.  We are on common ground here.

Content types are a new concept for the relational developer.  The content types that have the word ‘key’ in them have parallels to our familiar primary / foreign key concepts (KEY, KEY TIME, KEY SEQUENCE).  Let’s ignore them for a moment and concentrate on what remains:  DISCRETE, CONTINUOUS, and DISCRETIZED.

DISCRETE is a categorical value such as Male or Female, Married or Single, Between 32 and 37 years old, between 38 and 42 years old.  In the context of a content type it means the data is already presented that way.  The DISCRETIZED keyword just tells the engine that we want it to do the categorization.  The default is a 5 buckets approach.  If there is not enough data to support that, the engine will automagically try fewer.If that fails it takes a clustering approach.

CONTINUOUS values are familiar to most of us.  It means a numerical value.  Age columns that include values such as: 32, 37, 87, 17.  Monetary columns including amounts such as $1.42, $200.76, $432.45, etc.

image

MINING STRUCTURES

Before we can create any mining structures we have to create a new DMX query window.  This is accomplished by connecting to any Analysis Services database using SQL Server Management Studio, right clicking on SSAS instance and selecting: New Query, DMX.

image

All of the examples in this post will be based on two mining structures.  The first, simply called Customers, is our case level structure.  Don’t worry about what that means for now, let’s focus on the syntax.

CREATE MINING STRUCTURE [Customers] (
	 CustomerKey	LONG	KEY
	,CustomerName	TEXT	DISCRETE
	,Gender		TEXT	DISCRETE
	,Age		LONG	CONTINUOUS
	,AgeDisc	LONG	DISCRETIZED(EQUAL_AREAS,4)
	,MaritalStatus	TEXT	DISCRETE
	,Education	TEXT	DISCRETE
	);

We have a column name, a data type, and a content type very much like the SQL syntax for creating a table.

The possible values for the various columns are: gender – Male or Female, Age – 17, 32,64,42, etc., AgeDisc – 17-22, 23-28, 29-40, etc., MaritalStatus – Married or Single, Education: Bachelors, High School, Partial College, Graduates, etc.

CREATE MINING STRUCTURE [CustomersNested] (
     CustomerKey        LONG    KEY
    ,CustomerName        TEXT    DISCRETE
    ,Gender                TEXT    DISCRETE
    ,Age                LONG    CONTINUOUS
    ,AgeDisc            LONG    DISCRETIZED(EQUAL_AREAS,4)
    ,MaritalStatus        TEXT    DISCRETE
    ,Education            TEXT    DISCRETE
    ,Purchases            TABLE (
         Product            TEXT    KEY
        ,Quantity            LONG    CONTINUOUS
        ,Discounted            BOOLEAN    DISCRETE
        )
    ) 
    WITH HOLDOUT (30 PERCENT OR 10000 CASES) REPEATABLE (42);

This second mining structure is our nested table structure.  The nested table is a like a table that sits in a column within another table.  In this example, the column PURCHASES has a data type of TABLE.  It has a key column, the product name, a quantity column, and a discounted field.  Nested tables are like mining structure itself as they have a name and column list.

The HOLDOUT keyword will randomly select a certain percentage or number of records to be set aside for testing the model.

The REPEATABLE keyword will cause the same records to be set aside for testing the model each time it is populated if it used with a non 0 integer.  This is useful for testing the behavioral consistency of your scenarios.

Simple Simon was a Singleton

image

Here is an example of a record that is either being used to train a mining model within our nested table example, or as a singleton query to perform a prediction upon.  Notice the nested table is sort of a table in and of itself embedded in a column.  A singleton query is when you send one row to a model to make a prediction.  They are often used in real-time prediction engines.

image

MINING MODELS

ALTER MINING STRUCTURE [Customers]
    ADD MINING MODEL [ClusteredCustomers]
    USING Microsoft_Clustering;

DROP MINING MODEL [ClusteredCustomers];

Here we have simplest mining model money can buy.  This particular model would only be useful for clustering algorithms.  Notice there aren’t any column names in the mining model definition. If no columns are specified, all columns are considered inputs.

The USING keyword designates the algorithm your model will use. Understand that all algorithms use the same structures and model declarations and the same query language.

One thing I found out the hard way is that the more mining models your structure has, the longer it takes to populate your structure. This makes sense if you think about it, because it has to train so many models. If you go through the examples on your own, I’d recommend that you issue the drop statements or your structures will take forever to populate and some examples simply won’t work.

ALTER MINING STRUCTURE [Customers]
    ADD MINING MODEL [PredictMaritalStatus](
         CustomerKey
        ,Gender
        ,Age //NOTICE WE ARE USING THE CONTINUOUS VERSION OF AGE
        ,Education
        ,MaritalStatus PREDICT
        )
    USING Microsoft_Decision_Trees;

DROP MINING MODEL [PredictMaritalStatus];

This model is against our case level table structure, we are using the continuous version of the age column, and the model is using a decision tree algorithm.  The best way I can describe a case level structure is any structure without nested tables.

The word PREDICT in the model above is called a usage flag.  A usage flag has three states: PREDICT, PREDICT_ONLY, and NULL (the absence of a usage flag.)

  1. PREDICT means the columns is both an input and an output
  2. PREDICT_ONLY means the column is an output only.
  3. A column that is missing a usage flag is considered an input only.

We can only perform predictions against columns that are an output.  Input columns are fed into the model and used to make predictions on the output columns.

ALTER MINING STRUCTURE Customers
ADD MINING MODEL PredictMaritalStatusBayes(
     CustomerKey
    ,CustomerName
    ,Gender
    ,AgeDisc //NOTICE WE ARE USING THE DISCRETE VERSION OF AGE
    ,Education
    ,MaritalStatus PREDICT
    )
USING Microsoft_Naive_Bayes;

DROP MINING MODEL PredictMaritalStatusBayes;

In this example notice we are using the discrete version of the age column and the model is using the Naïve Bayes algorithm.  The algorithms Microsoft Naive Bayes and Microsoft Association Rules support categorical analysis and do not support continuous types.  The Microsoft Linear Regression algorithm accepts only continuous data.

ALTER MINING STRUCTURE CustomersNested
ADD MINING MODEL PredictMaritalStatusNestedTrees(
     CustomerKey
    ,Gender         //INPUT ONLY B/C USAGE FLAG IS ABSENT
    ,AgeDisc AS Age //INPUT ONLY B/C USAGE FLAG IS ABSENT
    ,Education      //INPUT ONLY B/C USAGE FLAG IS ABSENT
    ,MaritalStatus PREDICT //INPUT AND OUTPUT
    ,Purchases (
         Product
        ,Quantity   //VALUE ATTRIBUTE
        ,Discounted //VALUE ATTRIBUTE
        )
    )
USING Microsoft_Decision_Trees(COMPLEXITY_PENALTY = .5);

This model can be described as a model used to predict a person’s marital status based on that person’s gender, age, education, marital status, and the quantities of their product purchases, and whether or not they were purchased on sale.

ALTER MINING STRUCTURE CustomersNested
    ADD MINING MODEL PredictPurchasesTrees(
         CustomerKey
        ,Gender
        ,AgeDisc AS Age
        ,Education
        ,MaritalStatus 
        ,Purchases PREDICT (    //INPUT AND OUTPUT
             Product                
            )
        )
    USING Microsoft_Decision_Trees;

This model predicts product purchases based on a person’s gender, age, education, marital status, and other products purchased.  Since the table column Purchases is marked PREDICT instead of PREDICT ONLY, each attribute in the nested table is both input and output. If it were marked PREDICT ONLY it would not take into account the other products purchased.  The table is marked with the usage flag not the key. The model predicts the value of an attribute.  The model predicts the rows that make up the nested table. This is why it is the table that accepts the usage flag in this example.

For a nested table it is acceptable and common for it to have only a single column that is the key. Nested tables without supporting columns are the most common and are often used for market basket analysis.

The existence of a nested table row can be inferred by the existence of a value in any non-key nested column. This model creates what are called valueless attributes.

Valueless attributes are those attributes that do not have a separate table column representing their value. They are simply EXISTING or MISSING. The value of the attributes for the single case of Simple Simon would simply be:

  1. Purchases: Tires, Existing
  2. Purchases: Fenders, Existing
  3. Purchases: Playing Cards, Existing
  4. Purchases: Clothes Pins, Existing

Theoretically, the contents of a case also contain all of the “Missing” attributes too. So for every product you didn’t purchase that would also be a value attribute.

Microsoft’s algorithms have been written to specifically handle variable length cases. This is a radical departure from other data mining packages that assume each case is identically large.

ALTER MINING STRUCTURE CustomersNested
ADD MINING MODEL PredictQuantity(
     CustomerKey
    ,Gender
    ,AgeDisc AS Age
    ,Education
    ,MaritalStatus 
    ,Purchases (            //INPUT ONLY NESTED TABLE
          Product
         ,Quantity PREDICT    //INPUT AND OUTPUT
        )
    )
USING Microsoft_Decision_Trees;    

DROP MINING MODEL PredictQuantity;

This is an example of a Nested Table without a Usage Flag.  This model predicts the quantity of products purchased based on gender, age, education, marital status, and the quantity of other purchased products. The nested table contains a predictable column that is an input and output as it is marked with the PREDICT usage flag.

A valueless attribute (existing or missing) is not necessary and is not created because the table is an input and includes a value column that is also an input.

So valueless attributes were created in the previous example because all the nested table had was product which was either existing or missing. In this case they are not created since the quantity field is there they are not needed.

Question: What can we do with Quantity that we can’t do w/ any other column?

ALTER MINING STRUCTURE CustomersNested
ADD MINING MODEL PredictOnlyTable(
     CustomerKey
    ,Gender
    ,AgeDisc AS Age
    ,Education
    ,MaritalStatus 
    ,Purchases PREDICT_ONLY (    //OUTPUT ONLY
          Product                
         ,Quantity                //INPUT ONLY
        )
    )
USING Microsoft_Decision_Trees;    

DROP MINING MODEL PredictOnlyTable;

Answer: Predict.

This model predicts what products are likely to be purchased based on gender, age, education, marital status, and the quantity of other items purchased. A valueless attribute is created for the outputs since in this case you can’t predict quantity.

ALTER MINING STRUCTURE CustomersNested
ADD MINING MODEL PredictOnlyTableQuantity(
     CustomerKey
    ,Gender
    ,AgeDisc AS Age
    ,Education
    ,MaritalStatus 
    ,Purchases PREDICT_ONLY (        //OUTPUT ONLY
          Product
         ,Quantity PREDICT_ONLY        //OUTPUT ONLY
        )
    )
USING Microsoft_Decision_Trees;

This model predicts the quantity of products purchased based only on gender, age, education, and marital status. This model matches the table value column’s usage flag with the table usage flag. Since there isn’t a discrepancy between the usage flags, valueless attributes are not created.

Another way to think of this… Valueless attributes are created whenever the usage flags of the table column cannot be match with the usage flags of any of the nested table’s value columns.

ALTER MINING STRUCTURE Customers
ADD MINING MODEL FilterByAge(
     CustomerKey
    ,Gender
    ,Age
    ,Education PREDICT
    ,MaritalStatus 
    )
USING Microsoft_Decision_Trees
WITH FILTER(AGE > 30);

DROP MINING MODEL FilterByAge;

Filters can include case-level and nested-level columns.

ALTER MINING STRUCTURE CustomersNested
ADD MINING MODEL FilterByBalls(
     CustomerKey
    ,Gender
    ,Age
    ,Education PREDICT
    ,MaritalStatus 
    )
USING Microsoft_Decision_Trees
WITH FILTER(EXISTS(SELECT * 
                   FROM Purchases 
                   WHERE Product = 'Bearing Ball' 
                   AND Discounted));

DROP MINING MODEL FilterByBalls;

Contents of nested tables can be filtered.

ALTER MINING STRUCTURE CustomersNested
ADD MINING MODEL FilterByNested(
     CustomerKey
    ,Gender
    ,Age
    ,Education
    ,MaritalStatus 
    ,Purchases PREDICT (
        Product
        ) WITH FILTER(NOT Discounted)
    ) 
USING Microsoft_Decision_Trees;

DROP MINING MODEL FilterByNested;

Also worth noting is that Columns referenced in the filter are structure columns and need not be part of the model definition.

image

TRAINING

Once a mining model has been trained or processed, it contains algorithmic patterns derived from the data. Models are used against new data to perform predictions on any output columns defined during their creation. Patterns are referred to as the model content.

TRAINING METHODS

  • OPENQUERY
  • SHAPE
  • Other DMX Queries
  • MDX
  • Stored Procedures
  • Row-set parameters as the source data query for INSERT INTO

This blog post will cover the first two, OPENQUERY and SHAPE.

What do you get when you combine DMX w/ MDX in your company’s data mining project? Job security.

Data Sources

  • You can use the data source in the adventure works SSAS sample.
  • You can create one using BIDS as part of an Analysis Service project and deploy
  • You can use the AssProcs assembly.

AssProcs Detailed Instructions:
http://marktab.net/datamining/2010/07/10/microsoft-decision-trees-algorithm/

AssProcs Download:  Appendix B
http://www.wiley.com/WileyCDA/WileyTitle/productCd-0470277742,descCd-DOWNLOAD.html

CALL ASSprocs.CreateDataSource(
     'Adventure Works DW'              
    ,'Provider=SQLNCLI10.1;Data Source=localhost
      ;Integrated Security=SSPI
      ;Initial Catalog=AdventureWorksDW2008R2'
    ,'ImpersonateCurrentUser','','');

TRAINING OUR case level structure using OPENQUERY

INSERT INTO MINING STRUCTURE Customers (
         CustomerKey
        ,CustomerName
        ,Gender
        ,Age
        ,MaritalStatus
        ,Education
        )
        OPENQUERY(
             [Adventure Works DW]
            ,'Select 
                     CustomerKey
                    ,LastName 
                    ,Gender
                    ,DATEDIFF(YEAR,BirthDate,GETDATE())
                    ,MaritalStatus
                    ,EnglishEducation
              FROM dbo.DimCustomer'
              );

Random number Example

CREATE MINING STRUCTURE [CustomersRandom] (
     CustomerKey        LONG    KEY
    ,CustomerName        TEXT    DISCRETE
    ,Gender                TEXT    DISCRETE
    ,Age                LONG    CONTINUOUS
    ,MaritalStatus        TEXT    DISCRETE
    ,Education            TEXT    DISCRETE
    ,Random                DOUBLE    CONTINUOUS
    );

INSERT INTO MINING STRUCTURE CustomersRandom (
         CustomerKey
        ,CustomerName
        ,Gender
        ,Age
        ,MaritalStatus
        ,Education
        ,Random
        )
        OPENQUERY(
             [Adventure Works DW]
            ,'Select 
                     CustomerKey
                    ,LastName 
                    ,Gender
                    ,DATEDIFF(YEAR,BirthDate,GETDATE())
                    ,MaritalStatus
                    ,EnglishEducation
                    ,CAST((ABS(CHECKSUM(
                     NEWID())) % 1000) AS FLOAT)
                     /1000 AS Random
              FROM dbo.DimCustomer'
              );

TRAINING OUR NESTED TABLE STRUCTURE

INSERT INTO MINING STRUCTURE CustomersNested (
     CustomerKey
    ,CustomerName
    ,Gender
    ,Age
    ,AgeDisc
    ,MaritalStatus
    ,Education
    ,Purchases(
         SKIP
        ,Product
        ,Quantity
        ,Discounted
        )
    )
SHAPE {  
        OPENQUERY(
             [Adventure Works DW]
            ,'Select 
                 CustomerKey1 = CustomerKey
                ,LastName 
                ,Gender
                ,DATEDIFF(YEAR,BirthDate,GETDATE())
                ,DATEDIFF(YEAR,BirthDate,GETDATE())
                ,MaritalStatus
                ,EnglishEducation
               FROM dbo.DimCustomer
               ORDER BY CustomerKey'
            )
        }
APPEND (
            {
            OPENQUERY(
                 [Adventure Works DW]
                ,'Select 
                     CustomerKey2 = f.CustomerKey
                    ,p.EnglishProductName
                    ,f.OrderQuantity
                    ,CASE 
                        WHEN f.DiscountAmount > 0 
                            THEN CAST(1 AS BIT)
                        ELSE CAST(0 AS BIT)
                     END
                  FROM dbo.FactInternetSales f
                  JOIN dbo.DimProduct p 
                    ON p.ProductKey = f.ProductKey
                  ORDER BY f.CustomerKey'
                  )
            } RELATE CustomerKey1 TO CustomerKey2
        ) AS Purchases;

The SHAPE syntax is used to form the flat table source data to the hierarchical nested representation required by the mining structure. The relationship between rows and record set is accomplished using the RELATE keyword which maps a column in the outer record set to a foreign key column in the nested record set. You can add as many record sets as desired simply by providing additional definitions. The record set must be ordered by the columns used to relate them. This is a key requirement (pun intended.)

SKIP indicates columns that exist in the source data that will not be used to fill the structure. It is primarily used in cases where you don’t have control over the columns returned.

image

PREDICTION

Prediction: Applying the patterns that were found in the data to estimate unknown information. The final goal in any data mining scenario, it provides the ultimate benefits of data collection and machine learning which can dramatically influence how business is conducted.

DMX simplifies these possibilities by using a consistent syntax for prediction across all of the various algorithms. It allows predictions to be scheduled and results stored in various formats including relational databases. Some are performed in real-time.

Examples Include:

  • Future values of a time series (How much will we earn next month?)
  • What other products a customer might be interested in purchasing
  • Likelihood of a customer switching to a competitor
  • Will the borrower repay the loan?
  • Does anything seem abnormal here?
  • What is the most effective advertisement to display to this customer?
  • How can I classify customers, products, or events?

QUERYING STRUCTURED DATA

Three Rules

  1. You can only select cases from models that support DRILLTHROUGH
  2. You can only see data the model can see (unfiltered)
  3. By default, you will only see columns used with the model*

*There is a way to get the additional columns

Three Can Dos

  1. You can query column contents to see discrete values or continuous range values used in those columns.
  2. You can query the model content to explore data patterns discovered by the algorithms.
  3. WITH DRILLTHROUGH you can see how training data cases reinforce the validity of the patterns that were found.

PREDICTION JOIN

image

Consider a mining model to be a table containing all possible combinations of input and output variables. Now imagine you are doing a traditional SQL join to a theoretical table that will determine a prediction result. Such a theoretical table would be not be pragmatic considering the number of possible combinations within nested tables and impossible once you consider continuous columns. Models contain patterns learned from data in a compressed format allowing for efficient execution of predictions.

PREDICTION JOIN SYNTAX

image

Three Ways to Query Structured Data

  1. Select all cases
  2. Select cases as a flat record set
  3. Select only test cases
SELECT * 
FROM MINING STRUCTURE CustomersNested.CASES;
//SELECT ALL CASES

SELECT FLATTENED * 
FROM MINING STRUCTURE CustomersNested.CASES;      
//SELECT CASES AS A FLAT RECORD SET

SELECT * 
FROM MINING STRUCTURE CustomersNested.CASES 
WHERE IsTestCase();        
//SELECT ONLY TEST CASES

Select All Cases or All Test Cases Results

image

Select Cases as a Flat Record-set Results

image

WITH DRILLTHROUGH Clustering

ALTER MINING STRUCTURE CustomersNested
ADD MINING MODEL ClusterDrill(
     CustomerKey
    ,Gender
    ,Age
    ) USING Microsoft_Clustering
    WITH DRILLTHROUGH;

INSERT INTO ClusterDrill;

SELECT DISTINCT Gender 
FROM ClusterDrill;

SELECT DISTINCT RangeMin(Age),Age,RangeMax(Age)
FROM ClusterDrill;

SELECT * 
FROM ClusterDrill.CONTENT;

SELECT * 
FROM ClusterDrill.CASES 
WHERE IsInNode('001');

DROP MINING MODEL ClusterDrill;

imageimage

image

QUERY Predict Marital Status Using Naïve Bayes

DELETE FROM MINING STRUCTURE Customers;

ALTER MINING STRUCTURE Customers
    ADD MINING MODEL PredictMaritalStatusBayes(
         CustomerKey
        ,CustomerName
        ,Gender
        ,AgeDisc
        ,Education
        ,MaritalStatus PREDICT
        )
    USING Microsoft_Naive_Bayes;

INSERT INTO MINING STRUCTURE Customers (
         CustomerKey
        ,CustomerName
        ,Gender
        ,AgeDisc 
        ,MaritalStatus
        ,Education
        )
        OPENQUERY(
             [Adventure Works DW]
            ,'Select 
                     CustomerKey
                    ,LastName 
                    ,Gender
                    ,DATEDIFF(YEAR,BirthDate
                              ,GETDATE())
                    ,MaritalStatus
                    ,EnglishEducation
              FROM dbo.DimCustomer'
              );

SELECT 
     t.CustomerName
    ,Predict(MaritalStatus) 
        AS PredictedMaritalStatus
FROM PredictMaritalStatusBayes     
PREDICTION JOIN
OPENQUERY(
            [Adventure Works DW]
            ,'SELECT
                CustomerKey
                ,CustomerName = LastName
                ,Gender
                ,AgeDisc = DATEDIFF(YEAR,BirthDate
                                    ,GETDATE())
                ,MaritalStatus
                ,Education = EnglishEducation
              FROM dbo.DimCustomer'
        ) AS t
ON     PredictMaritalStatusBayes.AgeDisc = t.AgeDisc 
AND PredictMaritalStatusBayes.Education = t.Education
AND PredictMaritalStatusBayes.Gender = t.Gender;

image

DMX requires a fully qualified descriptor for all the column names in the mapping because joins are bound by column name rather than column order.

Predict Marital Status Nested Trees

SELECT 
    t.CustomerName
    ,Predict(MaritalStatus) AS PredictedMaritalStatus
FROM PredictMaritalStatusNestedTrees
PREDICTION JOIN
SHAPE { OPENQUERY([Adventure Works DW]
    ,'SELECT
         CustomerKey1 = CustomerKey,CustomerName = LastName
        ,Gender,Age = DATEDIFF(YEAR,BirthDate,GETDATE())
        ,Education = EnglishEducation,MaritalStatus
      FROM dbo.DimCustomer ORDER BY CustomerKey')
       } APPEND ( {    OPENQUERY([Adventure Works DW]
    ,'SELECT
         CustomerKey2 = f.CustomerKey
        ,Product = p.EnglishProductName
        ,Quantity = f.OrderQuantity,Discounted = CASE 
            WHEN f.DiscountAmount > 0 THEN CAST(1 AS BIT)
            ELSE CAST(0 AS BIT)
          END
      FROM dbo.FactInternetSales f
      JOIN dbo.DimProduct p ON p.ProductKey = f.ProductKey
      ORDER BY f.CustomerKey')
    } RELATE CustomerKey1 TO CustomerKey2
) AS Purchases AS t
ON PredictMaritalStatusNestedTrees.Age = t.Age
AND PredictMaritalStatusNestedTrees.Gender = t.Gender
and PredictMaritalStatusNestedTrees.Purchases.Product 
    = t.Purchases.Product
AND PredictMaritalStatusNestedTrees.Purchases.Quantity 
    = t.Purchases.Quantity
AND PredictMaritalStatusNestedTrees.Purchases.Discounted 
    = t.Purchases.Discounted
and PredictMaritalStatusNestedTrees.Education 
    = t.Education;

image

Singleton Queries

SELECT
    Predict(MaritalStatus) AS PredictedMaritalStatus
FROM PredictMaritalStatusBayes
NATURAL PREDICTION JOIN
(SELECT 'M' AS Gender
        ,35 AS Age
        ,'Graduate Degree' AS Education) AS t;    

SELECT
    Predict(MaritalStatus) AS PredictedMaritalStatus
FROM PredictMaritalStatusBayes
NATURAL PREDICTION JOIN
(SELECT 'F' AS Gender
        ,22 AS Age
        ,'High School' AS Education) AS t;

imageimage

NATURAL PREDICTION JOIN – matches columns from the source with the same names as input columns of the model.

SELECT
    Predict(MaritalStatus) AS PredictedMaritalStatus
FROM PredictMaritalStatusNestedTrees
NATURAL PREDICTION JOIN
(
    SELECT 'M' AS Gender
           ,35 AS Age
           ,'Graduate Degree' AS Education
           ,
    (        
        SELECT 'Touring Tire' AS Product
               ,2 AS QUANTITY 
        UNION
        SELECT 'Mountain-200 Silver
               ,38' AS Product
               ,1 AS QUANTITY 
        UNION
        SELECT 'Fender Set - Mountain' AS Product
               ,1 AS QUANTITY 
    ) AS Purchases
) AS t;

image

Degenerate Prediction– a prediction without source data and therefore has no prediction join clause. If 42% of my customers are married, the degenerate prediction for marital status would be single.

Predict Function Histogram

DELETE FROM MINING STRUCTURE Customers;
ALTER MINING STRUCTURE Customers
ADD MINING MODEL FilterByAge(
CustomerKey
,Gender,Age
,Education PREDICT
,MaritalStatus)
USING Microsoft_Decision_Trees
WITH FILTER(AGE > 30);

INSERT INTO MINING STRUCTURE Customers (
CustomerKey
,CustomerName
,Gender
,Age
,MaritalStatus
,Education)
OPENQUERY([Adventure Works DW]
,’Select
CustomerKey
,LastName
,Gender
,DATEDIFF(YEAR,BirthDate,GETDATE())
,MaritalStatus
,EnglishEducation
FROM dbo.DimCustomer’);

SELECT
‘Histogram’ AS Label
,PredictHistogram(Education) AS Hist
FROM FilterByAge;

SELECT FLATTENED
(SELECT $Probability
FROM PredictHistogram(Education)
WHERE Education = ‘Bachelors’)
FROM FilterByAge;

//RETURNS THE PROBABILITY ASSOCIATED W/
//THE VALUE ‘BACHELORS.’

SELECT FLATTENED
(SELECT Education,$Probability
FROM TopCount(PredictHistogram(Education)
,$Probability,5))
FROM FilterByAge;
//USES TOPCOUNT FUNCTION TO RETURN TOP 5 ROWS
//OF THE HISTOGRAM TABLE BASED ON PROBABILITY

DROP MINING MODEL FilterByAge;

image

imageimage

image

This is the most comprehensive prediction function for case-level columns. It returns a table w/ all the information available about the prediction of a scalar column.

Target – the predicted value.  Bachelors, partial college, and so on.
$Support – How many cases support this fact
$Probability – computed probability of a categorical output.  For continuous values represents the likelihood of a value being present.
$AdjustedProbability – modified probability used to boost likelihood of rare events, frequently used for predicting nested tables.
$Variance – of a continuous prediction.  0 for discrete predictions.
$StDev – of a continuous prediction.  0 for discrete.

When you’re dealing w/ functions like PredictHistogram that return tables, you can select rows from them like any other table.

DMX doesn’t allow TOP and ORDER BY clauses in sub-selects. TOPCOUNT AND BOTTOMCOUNT substitute for this functionality.

SELECT PredictHistogram(MaritalStatus)
FROM PredictMaritalStatusBayes;

SELECT
    MaritalStatus,PredictProbability(MaritalStatus)
FROM PredictMaritalStatusBayes;        

//Shortcuts for returning values such as the 
//probability and support:

//PredictProbability
//PredictSupport
//PredictAdjustedProbability
//PredictVariance
//PredictStdDev 

SELECT
    VBA![Log](PredictProbability(MaritalStatus)) 
        AS NaturalLog
FROM PredictMaritalStatusBayes;        
//EXAMPLE SHOWING HOW TO ACCESS SCIENTIFIC 
//FUNCTIONS FROM VBA SCRIPT LANGUAGE

First let’s look at the predict histogram function against the predictMaritalStatusBayes model.

image

Results from the PredictProbability function:

image

All functions return their values from the most likely row of the predictHistrogram record set without requiring messy sub-select syntax.

All of these functions also let you extract appropriate value for any row of the histogram.

Predict functions have Polymorphic behavior dependent on whether supplied case-level or column reference.

You don’t even need to use function when predicting a case-level column value. SELECT education = SELECT PREDICT(education) as long as education is a predictable column.

Results from the VBA![Log](PredictProbability) query:

image

Additional functions are shortcuts that give additional information about the prediction (support, likelihood, or the part of the model used for prediction.)

A reference of all DMX functions appears in Appendix B from the WROX download site mentioned previously in regards to the AssProcs download.

PredictAssociation

This function takes a nested table and returns a ranked list of results.  It is useful for recommendation predictions.

// PredictAssociation takes a nested table 
// and returns a ranked list of results

SELECT
    PredictAssociation(Purchases,3)
FROM PredictPurchasesTrees;
//top 3 items based on the default measure
//probability

SELECT
    PredictAssociation(Purchases,3,$AdjustedProbability)
FROM PredictPurchasesTrees;
//same but uses adjusted probability to rank the items

SELECT(SELECT * 
       FROM PredictAssociation(Purchases
                               ,INCLUDE_NODE_ID
                               ,INCLUDE_STATISTICS)
    WHERE $Probability > .1)
FROM PredictPurchasesTrees;        
//Shows all items that are recommended 
//with a probability of 20 percent or greater.  
//Also includes that statistics and node ID of 
//each prediction

imageimage

image

Adjusted Probability is useful in associative predictions because it penalizes items that are too popular in favor of less-popular items. If 90% of your customers want eggs, but for a particular customer the engine recommends milk with an 80% likelihood. The engine is telling you that for this customer the chance of them wanting milk is less than average.

//Predicting the values of columns in nested tables 
//instead of the contents of the tables themselves 

SELECT 
    (SELECT 
        Product
        ,Predict(Quantity)
    FROM Purchases
    )
FROM PredictOnlyTableQuantity;
//returns the predicted quantity for each 
//item assuming each item is purchased.

SELECT 
    (SELECT 
        Product
    FROM Purchases
    WHERE Predict(Quantity) > .42
    )
FROM PredictOnlyTableQuantity;    
//returns a list of all products with a 
//predicted quantity of greater than .42  

SELECT 
    (SELECT 
        Product
        ,(SELECT 
            $Probability
          FROM PredictHistogram(Quantity)
          WHERE Quantity = NULL)
    FROM Purchases
    )
FROM PredictOnlyTableQuantity;        
//returns a list of all possible products 
//for each case and in doubly nested 
//table returns the probability that a product 
//will NOT be purchased.

SELECT 
    (SELECT 
        Product
        ,Predict(Quantity)
    FROM Purchases
    WHERE PredictProbability(Quantity, NULL) < .8)
FROM PredictOnlyTableQuantity;    
//simplifies the previous example by using 
//a scalar function instead of the table 
//returning predictHistogram and returns the 
//predicted quantity of all products that are likely 
//to be recommended with a probability of at least 80%.
//i.e. determining the likelihood of it NOT being 
//recommended is less than 80%

imageimage

image

Predicting the values of columns in nested tables instead of the contents of the tables themselves is extremely useful; but, rare scenario.

Additional Mining Model Management Syntax

EXPORT MINING STRUCTURE Customers TO 'C:\c\Customers.abf'
WITH PASSWORD='mike';    

RENAME MINING STRUCTURE Customers TO Customer1;
DELETE FROM MINING STRUCTURE Customer1.CASES;

IMPORT FROM 'C:\c\Customers.abf' WITH PASSWORD='mike';

DROP MINING STRUCTURE customers;
DROP MINING STRUCTURE customersNested;
DROP MINING STRUCTURE customersRandom;

REFERENCE

image

Data Mining with Microsoft SQL Server 2008 is an excellent resource to learn more about data mining.  If you want to learn more about data mining with the Microsoft BI stack, this is a MUST read.

——————————————————————————-

This is a recording of my screen and voice during a presentation on the DMX Data Mining Extensions available in Microsoft SQL Server at SQL Saturday in Jacksonville, Florida on April 27th, 2013.

Data Mining with DMX from Mike Milligan on Vimeo.

——————————————————————————-

image

Generational Changes to Enterprise Data Warehouse Architecture

I had been on the job only three weeks before I began work on my secret plans.  It started out as just a simple BUS matrix; but, very quickly became a drill down BUS matrix of epic proportions.  I knew from the onset if my schemas were ever to hit a database I would need to be clever, patient, and a maybe even a little cunning.  I would have to keep my cards close to my chest, revealing my intentions only in bits and easily digestible bytes, sometimes as if they were not even of my own creation.

Prior to this gig I had been working as a hired gun, traveling the nation helping those in need of business intelligence consulting.  My particular specialty was the model.  It was always about the model in my views.  Actually, I prefer a physical model where the views merely reflect those structures; but, I digress.  I don’t know how many times I have seen a solution that relies on massive contortions of MDX to accomplish what would have been more scalable and elegant by simply cooking up another fact table to support that specific type of analysis.  Working as a consultant was a lot of fun and extremely challenging.  I am grateful for the opportunity I had to really make a difference on so many teams and projects.  However, after some time, the length of the journeys, the hardships of the trail, and my longing to be with my family led me to become a FTE at a billion dollar conglomeration of almost twenty individual companies.

Being a full time employee was a completely different role in my case.  No longer was I being asked to evaluate existing architectures and recommend best practice approaches as my primary function.  Initially, my priorities were mainly focused on immediate business needs and technical challenges.  In the first few months there were several challenging tasks outside of my comfort zone.  I couldn’t even estimate the distance in which I managed to herd an entire group of databases of an instance from one piece of hardware to another.  I built the structures for a 3 tier SharePoint farm (including PowerPivot) from my own bare hands, eight fingers, and two thumbs.  I implemented the processes and structures required to track prices and at least five ways to measure cost historically only after an elusive hunt for the mysterious “actual actual cost.”

Through it all I quietly kept working on my master plan.  I planted the seeds for it’s growth by mentioning what I refer to as “Kimballisms” (most folks refer to them as “Design Tips”) every chance I got.    I recited them like a preacher’s boy recites passages from the Bible, especially the dimensional modelers mantra:   “Ease of use, Query Performance!”  At the water cooler, I’d say things like, “If only we could drill across via some sort of conformed dimension.”  During the U.S. presidential election I expressed my hope that Ralph Kimball would run.  I pondered aloud how the world would be different if Kimball had designed all of Apple’s products.  “Load data at it’s most atomic level!” I’d mutter to myself as I wandered through the hallways.  It even got to the point that I’d attribute almost any words of wisdom to the man, “Kimball says not to go in the water for at least 30 minutes after eating.”

One day, an opening appeared.  My manager, whom I’ll refer to as Mr. Scratch, asked me to saddle up and produce a prototype PerformancePoint dashboard that would combine information about sales and collections.  I rassled up a prototype lickity split; but, there was a big problem.  The billing information was sitting pretty  on a SQL Server Analysis Services database; but, the sales roster was in a completely different database.  Furthermore, variations on similar dimensions complicated the matter such that the prototype had to have two different drop downs for basically the same information.  In one, the clerk number was in front of the storekeeper’s name and in the other it was after the name!  This was the perfect opportunity to talk about conformed dimensions with Mr. Scratch!  Much to my surprise, he took to the idea like a cat to fish so much so that I think he thought it was his own.  I don’t want to downplay the significance of this event.  Many existing reports and user generated solutions would become obsolete and have to be rewritten.  This was no trivial endeavor to suggest.  My plan was taking shape!  The seeds had sprouted and the roots were taking hold.

I couldn’t reveal my entire strategy all at once.  I had to bide my time and proceed carefully.  The existing data warehouse was created through the massive efforts of an extremely effective albeit small team.  They had gone through the painful trials and tribulations of understanding disparate legacy source systems and reverse engineering their business logic through painful iterations involving buckets of blood, sweat, and tears.  They had engaged the business and delivered analytic capabilities that had been unimaginable prior to the data warehouse.  The corporate buy-in was tremendous and the team had become the champions of the organization long before I had arrived.  They had bravely fought the good fight and won the hearts and minds of the entire corporation.  As can be expected, any suggestion about improvements had to be made with cautious candor.

It’s not hard to be tall when you get to stand on the shoulders of giants.  Their existing solution was better than at least three quarters of all the solutions I had ever seen.  They had avoided many of the common pitfalls and gotchas of data warehouse and OLAP cube development.  They were properly managing a type two slowly changing dimension.  Their robust solution was working, reliable, trusted, and respected by organization.  Important decisions were made every day based on the data provided by the data warehouse.

Eventually, it came time to share my proposed BUS matrix and drill down BUS matrices to the data warehouse manager.  Mr. Scratch played the devil’s advocate like a well worn fiddle.  We sat down to let him throw rocks at my model and had a rather heated debate on the subject of snowflake vs. star schema designs as he played his fiddle.  My first attempt was to cite Kimball design tip #105, “Snowflakes, Outriggers, and Bridges.”  In this tip modelers are encouraged “to handle many to one hierarchical relationships in a single dimension rather than snowflaking.”  Next, I referenced Chris Adamson’s Star Schema Central blog article Star vs. Snowflake.  Mr. Scratch was well versed in the debate strategies for refuting expert testimony and would have none of it.  That man sure can play a fiddle.  By the end of the discussion, I was convinced I would have to work something up that would measurably prove once and for all that the star schema was better performing than a snowflake.  I tried; but, lets just say that the results were inconclusive.  I personally still stand by Kimball and Adamson and I will accept their expert testimony.  For more on this subject, please listen to or read the transcript of the recent SQL Down Under podcast with guest Erin Welker, “Dimensional Modeling.”  I don’t want to beat a dead horse here so on with the story.

The next day, my manager went down to Georgia, and the discussion was eventually forgotten.  He never held me to a snowflake design pattern and we ended up with a strict star schema.  So here we are today, months after that initial design session and one week after an intense closed beta period.  Tomorrow marks the beginning of open beta and in three to four weeks the new data warehouse and cube replaces the old.

Looking over the summary of changes as I create a PowerPivot presentation for my manager I can reflect on all of the important design decisions that were made over the past weeks.  Most in business intelligence are familiar with Kimball’s concept of iterative changes and a dimensional lifecycle where incremental improvements are implemented to the data warehouse on a periodic repeating pattern.  These can include additional fact tables to existing data marts, changes to dimensions, and additional calculated measures.  Although I can’t remember exactly where, I do remember reading that a data warehouse isn’t considered mature until it has under gone its third generation.  Generational change occurs when the solution is effectively torn down and rebuilt from scratch.  This process uncovers unknown bugs and issues that may have gone undetected for years.  It forces one to rethink everything that has gone before at least in respect from the ODS to the data warehouse in this situation.

On the relational data warehouse side we went from 54 tables to 30.  Tables are differentiated by schemas that represent the business process they model.  Conformed dimensions are placed in a COMMON schema.  This was achieved primarily through the consolidation of related dimensions as attributes into a main dimension.

On the Analysis Services side we started with four databases containing one or two cubes each derived from one fact table per cube.  Now we will have one analysis services database, one cube, and six measure groups (three of which provide brand new analytic capabilities.)  We reduced the number of dimensions by about half primarily through consolidation.  We added several new dimensions and attributes and we have added time intelligence capabilities.

image

Old version – Snowflake Schema Design

image

New version – Star Schema Design

List of Significant Architectural Design Changes

  • Consolidated header/detail fact tables into one by allocating header values to the lower granularity
  • Conformed common dimensions
  • Consolidated dimensions
  • Added a new type of fact table with data that was not available before, an accumulating snapshot to track inventory levels
  • Added a new fact table to track prices and costs of products historically (and all of the components that make up those calculations I might add.)
  • Added a new fact table to track goals to actuals
  • Added attributes to represent discretization buckets
  • Proper cased the data elements themselves (everything was in all caps from the legacy systems and it’s much more purdy now)
  • Several new behavioral attributes to existing dimensions

Next on the Agenda

  • Introduce page level database compression
  • Introduce relational partitioning
  • Implement automatic analysis services partitioning mirroring the relational partitioning
  • Installation of an n-tier SharePoint 2013 farm
  • Creation of PerformancePoint 2013 dashboards
  • Schedulable forecasting using the DMX data mining query language and SSIS
  • Migration to SQL Server 2012 Reporting Services running in SharePoint integrated mode
  • Take advantage of new capabilities in PowerView and PowerPivot

I hope you’ve enjoyed this campfire tale.  I thought it might be good to share.  I’m tired of typing.  I’ve got blisters on my fingers!  My digits are weary and I can no longer see the forest through the trees  Imagine as I ride my old mustang off into the sunset playing a harmonica.

I’ll leave you with a mangled version of a poem inspired by Robert Frost.

Stopping by Woods on a Snowflaky Schema

Whose woods these are I think I know

His house is in the village though.

He will not see me stopping the server

to watch his woods fill up with snowflake schemas.

My little horse must think it BI

to stop without a SharePoint farm house near

Between the woods and frozen lake

the darkest evening of the fiscal year.

He gives his default wav a beep

To ask if I meant that last delete.

The only other sound’s the sweep

of easy wind and downy flake.

The woods are lovely dark and deep.

But I have promises to keep.

And miles to go before I sleep.

Miles to go before I sleep.

Business Intelligence Music Mix!

First EVER Business Intelligence Music Mix! SQL, OLAP, Data Mining, Computers, Geek Music. It doesn’t get any nerdier than this.  89 Tracks!  Includes great hits like:

Sequel – SQL
Data Mining – depth.charge
Binary – clammyhands
Byte – knightsofficial
bit – BIT 8
Sad – Programmer
President Kimball – Caesar
A French Winter – Database
DMX! – SameOIG
Press My Start Button, Please – TimothyPatrickBird
Software Check – nottall
Code Monkey – Jonathan Coulton
Bytes in Motion – DesertcoastMediaGroup
Microsoft Vista “Cuzco” – Steven Ray Allen
Microsoft Sam – Jesse Tippit
SSIS – JOERICH
Aggregation – wgramer
Estranged Apartmeant “Another Dimension (Integration)”
Slice & dics – Smokescreen dubstep
Different measure – d16group
Attribute – Philter
DSV – Mazz+X
Disaster Recover – Abnomally Sound Group
Jericho – DBA
CTRL ALT DLT – Tokiz’

… and many, many more!