Tuesday Apr 09, 2013

Big Data Analytics - Advanced Analytics in Oracle Database

That's the title of a new white paper we've just posted. From the executive summary:

Big data doesn’t only bring new data types and storage mechanisms, but new types of analysis as well. In the following pages we discuss the various ways to analyze big data to find patterns and relationships, make informed predictions, deliver actionable intelligence, and gain business insight from this steady influx of information. 

You can check it out here.

Thursday Jan 31, 2013

Using In-Database Analytics to Predict Fraud

Your data warehouse stores critical data telling you what is happening in your business and sometimes why it’s happening. But you can go beyond understanding why something went wrong. You can use past data to predict the future, correcting problems before they happen. In a recent survey that Oracle did of over 300 C level executives, 93% of them thought that their companies were losing an average of 14% of their total revenue because they couldn’t fully leverage the information they had already collected. One key way to do this (and you’ll hear more about this in a future survey) is to use predictive analytics. Let’s take a quick look at why and how.

Turkcell is a leading mobile phone provider in Turkey, with over 34 million subscribers. And like most mobile providers a majority of those subscribers use pre-paid accounts and pre-paid cards. Money launderers take advantage of this, and losses for this business are of the order of $5 for every $10,000. This may not seem like much, but with billions of transactions, this adds up to millions of dollars a year.

Like other companies, Turkcell examine huge quantities of data and build models that help it identify and ultimately predict and prevent fraudulent transactions.  Unlike many other companies, Turkcell does this analysis in its data warehouse. With 100 TB of compressed data – representing over a petabyte uncompressed – it would take a long time to move that data out of the warehouse and keep it up to date as new data arrived. And the window to stop the next fraudulent transaction might have already closed.

Oracle Advanced Analytics enables you to perform sophisticated predictive analytics inside a data warehouse. You can mine your data directly while it is inside the Oracle Database using either SQL or R language APIs or the Oracle Data Miner SQL Developer “work flow” GUI extension, depending on your need and existing skills. You build models for past behavior and use that to predict future behavior, improving your accuracy with time. And best of all, there’s no need to move the data around which takes time you might not have and also leaves you exposed to security risks. As Turkcell said “...we can analyze large volumes of customer data and call-data records easier and faster than with any other tool”

Monday Apr 09, 2012

Polyglot Analytics

Just like polyglot persistence, where a variety of data stores exist to handle a wide variety of data persistency requirements for different data management use cases, we are now seeing the emergence of a variety of BI and analytic solutions for delivering insight on all types of data. In the context of polyglot persistence, Oracle not only offers the number one relational database solution in the market today, we also deliver market leading NoSQL and In-Memory (TimesTen) as well as our Essbase OLAP engine where they serve as excellent solutions for handling a variety of persistency requirements. Of course, we've had a strong track record of also delivering enterprise class business intelligence and reporting solutions to the market and now have extended the BI and Analytics offering with Oracle Endeca Information Discovery, a data discovery solution that enables one to quickly explore all relevant data whether structured, semi-structured or unstructured in nature. And just like polyglot persistence solutions that compliment one another, we also see the same case for polyglot analytics. Our Business Intelligence solutions complement each other. In fact, they solve different problems and create different kinds of value. Why Polyglot Analytics exist. In a sense, its no different than programmers using different language to tackle different requirements and use cases. As an example, it would not be unusual for server side code to be written in Java while a web page is driven by Javascript or even Ruby. 

Business Intelligence clearly provides proven answers to known questions while out extensions, in this case, Data Discovery, it provides fast answers to new questions formulated by the business user. For example, when the business intelligence report says that warranty claims on the top-selling product went up 15% last month, the new questions are “What changed? What’s the root cause? What are customers saying about this? That exploration happens in a discovery app.

And the relationship goes both ways. Data Discovery creates new KPIs for the BI stack to deliver. For example, a consumer packaged goods company learned that preference for seemingly unrelated brands was highly correlated in certain customer segments. This came from a social media discovery app and suggested new KPIs they quickly pulled into their operational BI system.

Far from replacing their BI systems with data discovery, our customers have instead been able to get far MORE value out of their existing BI systems because they are able to re-focus them on solving the problems they are most effective for, and creating new practices around data discovery to get fast answers to new questions.

Just like NoSQL solutions solve different problems than relational databases, Data Discovery solves new problems that are different than traditional business intelligence and reporting:

The fact that data is available immediately creates demand for it. As more application, consumer, sensor, and mobile data is available to the business, the more the business wants to use that varied data for daily decisions that today get made on intuition and opinion.
In analytics, big variety is a bigger problem than big volume because it can’t be solved by more processing power alone. In addition, the cost and time required to combine diverse data together must come down.
The people making these decisions are experts in the business, not in writing SQL queries. They need a user experience that’s simple to learn and use and this is a core capability of Oracle Endeca Information Discovery. 

It combines structured and unstructured data from inside or outside the company. An enterprise solution must work with the full range of data that matters to an enterprise, including multiple structured sources with diverse schemas, like the vehicle warehouse and quality touch point application data; including unstructured data like the long-form text descriptions in the warranty claims; regardless of whether the data is inside the company, like the warehouse, or outside the company like the NHTSA claims or JD Power data.
It delivers in-memory performance, but is not memory bound. An enterprise solution must maintain fully interactive query response times even when the data is too big to fit in memory. Endeca realized this years ago when it combined search and browsing in eCommerce because search indices are often too big to fit in memory. Oracle EID is written for multi-core, multi-processor servers and uses proprietary optimization algorithms to exploit the full memory hierarchy from on-CPU cache all the way down to disk.  It is a solution for provisioning targeted discovery apps. It provides IT with a new capability to quickly deliver discovery apps wherever the business needs them. 

Over time, we will see strong adoption of Data Discovery applications that further compliment and augment Business Intelligence solutions and why we will continue to see polyglot analytics take hold. 
About

We're taking the pulse of the Business Intelligence and Analytics market based on our insights and our experiences with colleagues, customers,and partners.

Search

Categories
Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today