Mark Hudson

May 22 2013

Making Predictions with Microsoft Data Mining Tools – Part IV

After deciding which mining model is best, you need details to make predictions actionable.  Microsoft enables a couple of ways to access those details.

SQL Server Integration Services (SSIS) includes the Data Mining Query transformation which uses input data with an existing mining model to return a prediction.  You can automate predictions by including a Data Mining Query transformation in a SSIS package and returning predictions in a regular data flow.

To make the prediction exporting process easier to understand, I will use SQL Server Data Tools to manually export a batch of selected prediction details.

Read More

May 19 2013

Making Predictions with Microsoft Data Mining Tools – Part III

Once you have successfully processed your mining structure and mining models, it is time to view your results.  To begin viewing those results, simply click the Mining Model Viewer tab.

Within the Mining Model Viewer tab, select the mining model you wish to review.  Ideally you gave your mining models meaningful names so you can easily distinguish mining models from the names in the pull down list.  The actual screen shown depends on the algorithm used by your selected mining model.  For example, the screen print below shows the results of a Decision Tree mining model.

Read More

May 12 2013

Making Predictions with Microsoft Data Mining Tools – Part II

After claiming the easy use of Microsoft data mining tools, it is time to walk you through the process using Microsoft’s SQL Server Analysis Services (SSAS) and SQL Server Data Tools.

You do not need an OLAP cube, dimensional data model or data warehouse to start mining data.  SSAS’ data source view (DSV) is a semantic layer between your physical data source and data mining processes.  The DSV allows you to join tables to denormalize your source data and derive new values to transform your data.  Yes, the DSV can correct some data sins, but those corrections come with a cost.  The DSV extracts data from your physical data source every time you process a data mining model so DO NOT CONNECT YOUR DATA MINING DSV DIRECTLY TO YOUR TRANSACTIONAL DATABASE.  Ch

Read More

May 07 2013

Making Predictions with Microsoft Data Mining Tools – Part I

Accurate predictions remain the Holy Grail of every business intelligence initiative. Reporting data to show what definitely happened is beneficial. Analyzing data to determine why things happened the way they did is even better. But having the confidence to predict what will happen in time to take action and alter outcomes demonstrates the real power of data.

Despite the obvious value, organizations are just now achieving predictive analytics productivity according to Gartner.  Maybe the delay was a lack of data or the quality of the available data. Maybe the delay was a lack of tools or skills. Maybe the delay was simply a lack of understanding the tools available and the skills needed to leverage those tools.

Read More

Apr 26 2013

How to Turn off Slowly Changing Dimensions

Months after a successful dimensional data mart deployment, I was informed end users were unhappy.  The users were apparently dissatisfied with the slowly changing dimension component of the solution.  Improper product names continued to appear on reports because those products were named incorrectly when they were sold.  End users wanted the products sold to reflect the proper product names rather than some earlier erroneous version of the product names.

Read More

 

Disclaimer

The words and opinions expressed here are those of each article's respective author, and do not necessarily represent the views of CapTech Ventures.