Business Intelligence – Oracle

Archive for the ‘Data Mining’ Category

Oracle Data Miner 11g – Available for Download

Posted by Venkatakrishnan J on February 15, 2008

I came to know this via Charlie Berger, Oracle Data Mining Product Management. Oracle Data Miner UI for 11g is now out for download. One can download it from here. This contains some 2 new algorithms (Logistic Regression and Multiple Regression) and a number of enhancements. For more details i would recommend you to check this out here. I think its time for me to start blogging about some data mining features and how BI EE and data mining can actually work together. Charlie Berger had in fact offered me before to give a quick tutorial on the various data mining features(Thanks Charlie for that offer!!!, i believe i need it now!!!). Somehow, i never got around doing any of those before. Maybe its time for me to start working with Ramkumar Krishnan(his blog has some excellent technical details on ODM) and Charlie to blog about some interesting points of integration between BI EE and ODM.



Posted in All Posts, Data Mining | 4 Comments »

Oracle Data Mining and Business Intelligence

Posted by Venkatakrishnan J on July 25, 2007

I believe Oracle Data Mining is one of the most under utilized tools/database options in the BI space. There are so many Oracle Data Warehouse implementations out there but hardly few use Oracle Data Mining (but yes the market share is growing recently). Is there any reason for such a low adoption rate? Is it due to the lack of capabilities of Oracle Data Miner? Or is it due to the perceived notion that Data Mining is for statisticians and the like? Or is it because the management does not believe in the value of Data Mining?

I think it’s primarily due to the perceived notion rather than the lack of features because ODM indeed has a very rich set of functionality (a lot!!!). Of course statisticians can much appreciate the value of a data-mining tool. But again, as a person who has been involved in some DW implementations, I believe we understand the data better than anyone else. To use data mining Statistical knowledge would be an absolute plus but not mandatory. I am planning to put together a series of articles that would elucidate the various models that are used in Oracle Data Miner with examples. Its more of an education for me too since I myself would be getting my hands dirty for the first time with this tool. It is a bit too ambitious but let me see how far I can go. I will try to put together some relevant day-to-day examples as and when I get time (I will try as much as possible to deviate from the Beer bottle and Baby diaper example J). For more detailed analysis and inputs we always have the documentation and the ODM PM Marcos Campos’s blog. Actually the ODM PM is Charlie Berger. Thanks for correcting me Charlie. But some usual caveats here. Data Mining is a huge and a vast topic. I plan to just cover the models that might make sense for all those who use Business Intelligence. If you have any feedback do let me know since I think I am being too opinionated here!!!

Before we get into the details, lets go through an overview of datamining types. There are 2 types of datamining.

       1. Supervised Data Mining
       2. Un-Supervised Data Mining

Supervised is also called as Predictive data mining wherein the process is used to predict some outcomes. For example, a credit card company can use this to predict whether a customer is going to default on his payments.

Un-Supervised Datamining, also known as descriptive datamining involves finding out intrinsic patterns within the data. Again, how the pattern is identified depends on the model.

Following are some of the models supported by Oracle Data Miner.

Supervised Data Mining:

        1. Classification
        2. Regression
        3. Attribute Importance
        4. Anomaly Detection

Un-Supervised Data Mining:

        1. Clustering
        2. Association
        3. Feature Extraction

With that I would like to close this with some interesting articles on datamining

        1. Crime Detection
        2. Credit Card Fraud Detection

Posted in All Posts, Data Mining | 2 Comments »