Advertisment

High Performance Analytics

author-image
PCQ Bureau
New Update

With the turn of decade we saw how business analytics has become analogous to

business intelligence, and how every BI solution incorporates analytics

capabilities to help enterprises make better business decisions. The data

warehouses of organizations are growing at incredible rate, and as organizations

perform highly complex and interactive queries against large volume of data and

do granular level of data analysis against that to gain insights for better

business decisions. They do not want to rely on historic data, but also on the

current streaming data. That would help organizations to make more relevant

decisions which are based on information of prevailing business scenario. Hence,

to do real-time data analysis and avoiding latency in query executions,

organizations require high performance analytics.

Advertisment

Direct Hit!

Applies To: CIOs



USP: Learn about the new trends in analytics


Primary Link: None


Search Engine Keyword: high performance analytics

What is high performance analytics?



High performance analytics can be defined as the technology that enables

complex analysis of data, phenomenon or information using integrated

computational methods on massive amounts. That would support critical reasoning

leading to proper insights and understanding towards the context of applied

challenge. Such advanced analytics incorporates methodologies for answering

future-oriented, proactive and predictive questions, as well as streaming data;

so that consequent business decision is based on real-time questions about

what's going on now. Advanced analytics leverages the same core features of

typical analytics solution, i.e. reporting, dashboards, visualizations etc. but

takes the analytics power to several steps further. Such powerful analytics

capabilities would be required by enterprises where they generate bulk of data

and there is a need to analyze that in real-time. For instance, large investment

or trading houses that invest large amount on shares they purchase daily in

stock markets. These trading houses rely on algorithmic trading that allows

these investors to obtain best possible price. Depending on these methods, these

investors buy large quantity of stocks and hold those momentarily for minutes or

even seconds and then sell them off to gain benefit. For this they constantly

monitor the fluctuating stock prices and compare those with the specified

parameters and constraints along with past performance of that particular

company. With buying or selling decision to be taken instantaneously before the

price of stock changes require real-time analysis of the stock market data with

minimal latency required in analysis process, and for such purposes high

performance analytic solutions like Sybase RAP are required. Such high

performance analytics isn't required by every organization, but companies which

are in domains of banking or telecom, where data is in bulk and output has to be

in real-time require tools having capability for high performance analytics.

Need for advanced analytics



For an enterprise implementing a Business Intelligence solution is not a

destination to arrive at, but rather a journey to begin with. With each passing

milestone being the series of processes in which the enterprise incorporates

various business processes and data-silos to achieve seamless flow of

information across the enterprise and thus have a “one view” of information.

That means throughout the enterprise it acts as a single data pipeline that is

accessed by all business processes and departments. Enterprises need variety of

analytics capabilities for addressing diverse decisions that need to be made by

the range of users, varying from top-notch business heads to sales personnel who

all would be resorting to the BI solution for that purpose. With advances in

data collection and storage technologies, large data sources have become

ubiquitous with every enterprise. The enterprise does want to do data analysis

on that, and that too speedily. Whereas, the current process includes

data-cleansing, extractions, transformation and loading of data into the

data-warehouses, and then through data-mining granular information can be

obtained from that data. This results in latency in the data relevance, and also

as data volume is huge the query execution time is also high. This is a cause of

concern for organizations that are in financial dealing, stock trading etc.

where data relevance is momentarily.

Advertisment

High performance platforms



Since the enterprises are collecting terabytes of data on a daily basis, it

has become imperative that the data mining and machine learning techniques also

scale up to such proportions. Such scaling can be achieved through design of

faster algorithms for query analysis and through employment of parallelism. The

emerging processor architecture, like multi-core, have made it possible to

achieve performance gains through parallelism. To address to these scales of

information management and processing, have seen the emergence of platforms like

Hadoop from Apache and Microsoft's Dryad. These platforms can support

implementation and deployment of large-scale analytics, and can support

execution on the cloud. Thus, offering to draw out more information out of

available data with higher efficiency and lower costs for the enterprises.

Thanks to the advancements in processor technology, data-mining and data

storage methodologies and techniques, the challenge to build platforms for

supporting high performance analytics is no more technical. Rather, road-blocks

are faced in form of enterprises still having silos of business information and

lack of enterprise information management procedures to streamline data flow

across business processes. The ever-increasing volume of data from heterogeneous

data sources has pushed the need for high-performance solutions so that

organizations can gain proper insight and competitive advantage from all the

data they posses.

Sybase RAP — The Trading Edition Edition

Together, Sybase RAP and Sybase CEP provides a

platform for supporting high-velocity trading and risk management

for trading firms in capital markets.

Capital markets industry today is facing a challenge to have better

risk management after the global recession, as they have ever-growing

data volume and the high volatility of that data. The Sybase RAP - The

Trading Edition is the next-generation market analytics platform that

enables capital market firms to make better trading and portfolio

decisions with less risk through timelier, more comprehensive market

insight.

It provides a real-time cache and historical repository with

ability to load, store and analyze with time series data. While, it is

the Sybase CEP (Complex Event processing) engine that processes the

continuous incoming streams of market data with ultra-low latency

measured in milliseconds. It is an enterprise-class platform that

addresses the key technology capabilities required by today's capital

markets firms. For instance, in Wall Street trading firms rely on

algorithmic trading. Algorithmic trading also called as 'algo trading'

is based on use of computer programs for entering trading orders with

the computer algorithm deciding on aspects of the order such as the

timing, price, or quantity of the order, or in many cases initiating

the order without human intervention. A platform like Sybase RAP (Risk

Analytics Paltform) consists of RAPStore, which is the repository of

market events, and RAPCache that provides access to recent market

events within seconds.

The combination of Sybase RAP and CEP then can analyze through the

streaming data of shares from stock market and compare it with past

records to predict their immediate pricings and its risk analysis.

Then these firms can benefit a lot through holding stocks for minutes

and then selling them. This, intra-day stock trading would help them

to capitalize through situations such as the recent recessionary

times, where most trading firms lot a fortune. For such analysis where

the data that is analyzed holds relevance for few seconds or even

less, means systems should support analysis of high speed streaming

data with very low latency.

Advertisment

Social Network Analytics



Social network had been the buzzword for the past year from an enterprise

perspective. Enterprises were monitoring their brand's reputation in the online

space and taking necessary steps to maintain that. According to Forrester's

report, the year 2010 will emerge as the year when social network analysis will

attain to new frontier in advanced analytics, which would incorporate behavioral

and attitude analysis of individuals. For social media analysis organizations

would be able to listen to all the tweets and updates in Facebook and what

people or consumers are talking about their products, the kind of interactions

going about their services or brands in forums. Then, this stream of event

information would be brought to data warehouses for analysis, and compared that

with historical information on those customers or cluster of people. This ways

an organization would be able to sense how people would respond to new offers,

campaigns or products, and what would be the best manner to offer those to their

consumers.

It's a predictive model, and involves complex event processing to analyze the

behaviors of networks of people on Internet. Even, they can change or modify

their marketing campaigns on the fly by getting users' views about it online and

prevent the campaign from failure. Thus, Internet being a huge source, it will

enhance an enterprise if its social network analysis supports real-time customer

segmentation, target marketing, churn analysis, and anti-fraud features, which

can be achieved through high-performance analytics tools.

Enhanced analytics solutions



The information does't necessarily has to be in a textual format, it can be

in form of images, audio, video etc. Some of the analytics solutions that are

coming up have phonetic search and recognition engines. For example, TelStrat's

Engage Analyze can index and mine audio words and phrases buried in the calls.

Advertisment

Unlike other speech to text approaches, the phonetic speech search is quicker

and can index content at 60-80 times faster. And these new systems learn on

their own as they record different phonetic for words. In a scenario where

customer calls have to be attended to, this type of system can comprehend user's

query and on its own find the most accurate solution for that to the caller.

Thus, automating the minor trouble-shooting problems of the users/callers.

There are other analytic engines that support multiple language support as

well. UTOPY's SpeechMiner can support multiple languages like Italian, Chinese,

Japanese, German besides English.

It allows processing of unlimited customer calls from across geographies.

Such high performance analytics is being deployed to reduce costs as interaction

costs with self-service are a fraction when compared to a person attending those

customer calls.

Advertisment