Thursday, November 19, 2015

My presentations at UKOUG 2015

The annual conferences of the UKOUG are coming up soon and I just wanted to give you a quick overview of the 3 presentations I will be giving at these conferences in Birmingham between Sunday 6th and Wednesday 9th December.

I'll be making references to 50 Shades of Gray and Lord of the Rings in some of these presentations. Come along to find out why.

It all kicks off on the Sunday (12:30-14:20) with an almost 2 hour presentation titled 'Predictive Analytics in Oracle: Mining the Gold & Creating Valuable Products'. We have being hearing about Predictive Analytics for a few years now. But most people are still unsure of what it really means and how it could be used in their company. This presentations aims to simplify what Predictive Analytics means and to illustrate how various companies from a wide range of industries have been using Predictive Analytics to gain deeper, valuable and actionable insights of their data. A number of case studies will be presented, based on some of my recent projects, illustrating how predictive analytics has been used to generate significant benefits for each company. These case studies will include examples from Banking, Telecommunications, Financial Fraud Detection, Retail and Merchandising. The second half of this seminar will look at what functionality is available from Oracle that allows you to build and deploy Predictive Analytics in your applications. We will at what you can do with SQL, PL/SQL, Oracle Data Miner, Oracle R Enterprise and Real-Time Decisions. We will then look at how you can build these features and tools into your front end applications. In the final section of the seminar we will look at what you need to do to get started with Predictive Analytics and how you can avoid some of the typical pitfalls.

On Monday 7th, I have a presentation between 9:00 and 9:50 titled 'Is Oracle SQL the Best Language for Statistics?'. Did you know that Oracle comes with over 280+ statistical functions? and that these statistical functions are available in all versions of the Database? Most people do not seem to know this.When we hear about people performing statistical analytics we can hear them talking about Excel and R. But what if we could do statistical analysis in the database, without having to extract any data onto client machines. That would be really good and just think of the possible data security issues with using Excel, R and other tools.This presentation will explore the various statistical areas available in SQL and I will give a number of demonstrations on some of the more commonly used statistical techniques. With the introduction of Big Data SQL we can now use these same 280+ Oracle statistical functions on all our data including Hadoop and NoSQL. We can also greatly expand our statistical capabilities by using Oracle R Enterprise using the embedded capabilities in SQL.

My final presentation will on Wednesday (14:30-15:20) titled 'Automating Predictive Analytics in Your Applications'. What does your Oracle DBA and Oracle Developer need to know to implement Data Mining? How can the Oracle DBA and Oracle Developer use Data Mining. This presentation will look at what the Oracle DBA and Oracle Developer needs to know to deploying, implementing and integrating the predictive models produced by the Data Miner. These predictive models can be used in multiple databases and across their many applications. Examples will be given of the typical tasks data mining task for the DBA and for the Oracle Developer. These examples will include SQL, PL/SQL and embedded Oracle R Enterprise. I will show how easy it is to build automated Predictive Analytics into, not only your Dashboards, but your everyday applications. Here is a short video that gives an overview of my presentations for TECH15.



Sunday, November 15, 2015

Error when trying to use GLM in ORE

If you have tried to use the ore.odmGLM function in ORE and some other functions in ORE and you are using the Oracle 12.1.0.2 Database you will be getting an error something like the following in R.

> GLMmodel <- ore.odmGLM(AFFINITY_CARD ~., ANALYTIC_RECORD)
Error in .oci.GetQuery(conn, statement, data = data, prefetch = prefetch,  : 
  ORA-29400: data cartridge error
ORA-40024: solution to least squares problem could not be found
ORA-06512: at "SYS.DBMS_DATA_MINING", line 2153
ORA-06512: at line 1

or you get something like the following when using ore.glm

Error in .External(C_pt, q, df, lower.tail, log.p) :
  NULL value passed as symbol address

Similarly if you are using these functions with embedded R execution when the R code is warpped in SQL, you will see similar errors.

What you need to do is to download and install the stand alond Patch 20173897.

NewImage

Alternatively this patch is included in the latest Oracle 12.1.2 Database patch.

Thursday, November 5, 2015

Slide from my OOW15 Presentations

At Oracle Open World (OOW15) I gave 2 presentations on the Sunday during the Oracle User Group Forum. The slides are now available for download from the Oracle Open World website.

Go get them now!

More Than Another 12 on Oracle Database 12c [UGF3190]

During this sessions I was one of 16 presenters talking about various features in the Oracle Database. All of the presenters where from the EOUC region.

Real Business Value from Big Data and Advanced Analytics [UGF4519]

I co-presented with Antony Heljula from Peak Indicators. During this presentation we talked about some of the Advanced Analytics projects we have worked on over the past 18-24 months. We also announced a new Analytics-as-a-Service offering.

The slides are also available for most of the other Oracle Open World Presentations and these can be accessed here. Just go search for the topic you are interested in.

Check out my previous blog post that summarises just a small part of what I got up to at OOW15.