Yes, you heard me correctly. Oracle has built-in technology that enables end-users and applications to perform advanced analytics without extracting data from the database. This functionality is sold as an option to the Enterprise Edition of the Oracle database - the Oracle Data Mining option. This post is the first of a series of posts that will describe Oracle's predictive analytics offering.
First, a short history lesson. Oracle acquired the assets of Thinking Machines Corporation in 1999. At the time, Thinking Machines was a developer of advanced data mining software, with a strong emphasis on parallel computing (which was a natural extension from their days as a hardware company developing the Connection Machine). After the acquisition, Oracle continued to sell Thinking Machine's existing product, Darwin, but also embarked on a decade-long strategy of embedding the core technology within the kernel of the Oracle database.
Flash-forward to today. Oracle Data Mining has a large assortment of algorithms as well as a deployment environment that is a native component of Oracle's SQL language and execution engine.
Oracle Data Mining supports the following data mining functional areas:
Oracle Data Mining algorithms include (abbreviations provided below to simplify my life in future posts):
- (AP) Apriori
- (DT) Decision Tree
- (GLM) Generalized Linear Models
- (KM) k-Means
- (MDL) Minimum Description Length
- (NB) Naive Bayes
- (NMF) Non-negative Matrix Factorization
- (OC) O-Cluster
- (SVM) Support Vector Machines
The purpose of this blog is to demonstrate the benefits of performing analytics directly in the database by exploring the power of Oracle Data Mining. Performance, manageability, security, reliability - and everything else the Oracle database has achieved over the last three decades - is now extended to the realm of analytics.