Vous êtes sur la page 1sur 4

Pivotal Data Science Labs is the new name for Greenplum analytics lab.

Data Sheet

Pivotal Data Science Labs

Expert-led discovery for actionable business insight

Overview
At-a-glance

Onsite assistance delivered by experienced data scientists


to help:

Solve critical problems using advanced analytics and


deliver actionable insights

Advance new methodologies within the full range of


the Pivotal UAPs analytical capabilities

Learn and apply Magnetic, Agile, and Deep design


philosophies to analytics

Key benefits

Rapid insights into critical business questions

Training for key analysts

Conversion of existing models

A roadmap for on-going analytics including sample


models, tools methods, and best practices

An opportunity to guide Pivotal analytics development

Accelerating Analytics in
a Big Data World
Acquiring and developing advanced analtyics skills is a key
priority for many organizations. Faced with limited staffing and
skills, while demand for talented analytics and data science
professionals is on the rise, many organizations struggle to grow
their abilities and deliver results quickly.
To help both existing and prospective Pivotal users develop
actionable insights for the business and grow their skill base
more rapidly, Pivotal has assembled a team of experienced data
scientists that is available for analytics-focused engagements.
Through our Data Science Labs, Pivotals data science team can
help accelerate analytical skill development and kick-start your
ability to deliver immediate value to the business.

Pivotal Data Science Labs


A Data Science Lab is a package of services, technology, or
training delivered by Pivotals team of leading data scientists.
During the engagement, your analytics stakeholders and data
platform leadership work in partnership with Pivotals team of
statisticians and modelers to solve real business problems using
Big Data advanced analytics.
Data Science Labs are collaborative projects that can include:

goPivotal.com

Studies to build analytical insight regarding key business


problems

Development of analytics roadmaps

Data asset study and validation

Scoping and initial development of analytics application


and algorithms

Alignment of analytics goals with business needs

Data Sheet Pivotal Data Science Labs

WORLD-CLASS DATA SCIENTISTS GUIDE


YOUR SUCCESS
The Data Science Labs are delivered by Pivotals team of leading
data scientists. Pivotals data scientists partner with your analysts,
data platform administrators, and business leadership to crack
your top business challenges and opportunities over a project
duration of your choosing. In each Lab, our data scientists
identify and, in some cases, implement the appropriate analytical
methods on massive datasets using the full range of capabilities
of the Pivotal Unified Analytics Platform (UAP).

HOW A PIVOTAL DATA SCIENCE LAB WORKS


Identify goals for the project that are critical to the business
and amenable to advanced analytics; agree on priorities and a
timeline.

Build a set of detailed requirements and exit criteria.

Assemble the Pivotal team to execute: in addition to the


data scientist, we provide project management, architectural
oversight, and data migration/loading support for the project.

Employ an iterative approach while pursuing clear project


goals and milestones, encouraging a process of discovery
based on what the data reveals, all targeted to contribute to
project goals.

Work closely with IT, paying close attention to security,


permissions, protocols, and testing procedures.

CHOICE OF MODELING AND


STATISTICS TOOLS

ALPINE Data Labs


Pivotal has worked with Alpine Data Labs to develop an intuitive
graphical workflow model builder for data mining, seamlessly
integrated with the Pivotal Greenplum Database. Alpine Miner
provides statistical transformation and modeling methods for
data analysis, modeling, and scoring with which analysts can
flexibly and efficiently conduct end-to-end knowledge discovery
and predictive analysis. Alpine Miner operates directly on data
where it resides, regardless of the number of independent
variables or complexity of the data types.
PMML
PMML, or Predictive Modeling Markup Language, while not a
tool itself, provides a method for exchanging models between
model development and model execution environments. PMML
can help to optimize development processes and Pivotal data
scientists can help you begin to leverage PMML to speed analytics
development.

TECHNIQUES FOR ACCELERATING


AGILE ANALYTICS
Optimizing analytics on Big Data requires new techniques to
harness the massively-parallel computational capabilities of UAP.
During a Pivotal Data Science Lab, we can help your team apply a
variety of the following new techniques quickly.
Embedded Analytic Functions
Pivotal is dedicated to bringing the power of parallelism to
commonly used modeling and analytics functions, and supports
many of these within UAP including matrix operations, multiple
linear regressions, and Bayesian statistics.

SQL
Python R
C Java
Perl

Your analysts will likely employ a wide variety of analytics,


development, and business intelligence tools. Pivotal Data
Science Labs are unique in the broad array of these technologies
and approaches that our data scientists can support. Some of
the major analytical technologies supported are as follows:
SAS
During Pivotal Data Science Labs, our analysts can help you
leverage and improve analytics initiatives for your SAS users.
In addition to analytics and modeling in SAS, we can also help
you run models directly in Pivotal UAP using the SAS InDatabase Scoring Accelerator for Pivotal and the new SAS High
Performance Analytics (HPA) for Pivotal.

MADLIB
The analytics team at Pivotal is actively contributing to an opensource library of advanced analytics functions in cooperation
with the University of California at Berkeley designed to run on
MPP platforms. These functions are available at no cost as part of
the MADLib analytics library, including documentation and source
code from which users can customize the algorithms.
MapReduce
MapReduce has proven to be a powerful platform for Big Data
analytics by Internet leaders including Google and Yahoo!. Pivotal
UAP uniquely supports MapReduce in both Pivotal HD and Pivotal
Greenplum Database, giving your analysts freedom to choose the
right tool and environment for each job. Pivotals data scientists
can help you effectively apply MapReduce and jumpstart its use in
Pivotal UAP.

Data Sheet Pivotal Data Science Labs

Hbase, Pig, and Hive


The Hadoop community is rapidly augmenting MapReduce with
new higher-level tools, including the Apache Foundations Hbase,
Hive, and Pig toolsets. Each is included in Pivotal UAP, and can be
the utilized during an Data Science Lab engagement to address
new analytics challenges.
SQL Analytics
SQL, and more specifically SQL 2003 OLAP functions, are
commonly used in analytic environments. Our data scientists can
help your team tune and optimize SQL to accelerate your SQL
analytics in the massively-parallel environment of Pivotal UAP.
Custom Analytical Algorithms
For many, adapting existing algorithms to run in a massivelyparallel environment can vastly increase analytical agility. For
computer scientists, well-known procedural languages including
Java, C, Python, and Perl can be used to create algorithms
that harness the parallel computational power of UAP. For
statisticians, the R programming langauge can also be used to
parallelize existing and create new analtyical algorithms.
During Data Science Labs, Pivotal data scientists can help you
exploit any of these procedural languages for flexibility and
performance, while shortening the time-to-value for highperformance agile analytics.

DATA SCIENCE LAB PACKAGES


Pivotal Data Science Labs are available in a range of
engagement durations and deliverables:
Lab Primer
(One-Day Workshop)

Lab 100
(Analytics Bundle)

Lab 600
(Six-Week Lab)

Lab 1200
(12-Week Lab)

Analytics Roadmap

Onsite MPP
Analytics Training

Analytics Roadmap

Analytics Roadmap

Analytics Toolkit

Prof. Services on
Pivotal UAP

Prof. Services on
Pivotal UAP

Quick Insight
(Two weeks)

Ready-to-deploy
Model(s)

Ready-to-deploy
Model(s)

Prioritzed
Opportunities
Architectural
Recommendations

LAB PRIMER (One-DAY WORKSHOP)


A Lab Primer is a one-day moderated session bringing together
your data and business leadership with Pivotals data scientists
and architects to review the existing data platforms and business
goals. Through the day, teams discuss opportunities and
approaches to apply advanced analytics, and to chart a concrete
path toward making this a reality.
The result is an analytic roadmap, with a step-by-step guide for
an analytics-based approach to solving one to three business
opportunities, as well as tactical and strategic recommendations

for data, process, and platform enhancements to enable best-inclass analytical performance.
This option is appropriate for companies that are new to
exploring analytics on massively-parallel-programming (MPP)
platforms, those that are in the beginning stages of elevating
analytics as a mission-critical business function, or those
suspecting they are under-leveraging valuable data assets.
LAB 100 (ANALYTICS BUNDLE)
Lab 100 engagements are typically two weeks long, working
with your data and analytics team to assist with introducing or
optimizing your Pivotal UAP analytics environment. In addition,
our data scientists work closely with your team to ensure that
users are fully equipped with the tools to leverage the advanced
analytics capabilities of Pivotal.
The result is onsite analytics training with the Pivotal UAP
targeting your future in-house data scientists, a review of
languages and tools such as SQL, R, MapReduce, MADLib, SAS,
and Alpine Miner (a GUI-based statistical package optimized for
Pivotal) and the presentation of a business insight by our data
science team.
This option is appropriate for new or existing Pivotal
customers who are interested in jumpstarting their advanced
analytics efforts to maximize the performance of their Pivotal
environments and the value theyre extracting from their data
assets.
ANALTYICS LAB 600
An Analytics Lab 600 is a six-week model-development
engagement focusing on solving a top business challenge or on
discovering a key insight that can be further operationalized to
address marketing or product goals.
The result of the Lab 600 is typically a QAd, ready-to-deploy
model or set of models that are tuned to optimally perform in
a Pivotal UAP environment.
This option is appropriate for companies with a known business
challenge that can benefit from the brief injection of additional
analytics experience. Examples of possible business challenges
addressable in a six-week timeframe include: customer
segmentations, affinity models, and experimentation frameworks.
ANALYTICS LAB 1200
Analytics Lab 1200 is a 12-week Pivotal UAP-based Analtyics Lab
engagement, focused on solving a top business challenge deemed
more complex than would be tractable in a Lab 600 engagement.
As with the Lab 600, the focus is usually upon discovering one or
more key insights that can be further operationalized to address
3

Data Sheet Pivotal Data Science Labs

marketing or product goals. As with a Lab 600, our data science


team works with you to address your analytics challenges, work
hands-on with your data, and deliver actionable insights for your
business or organization.
The result of a Lab 1200 is a QAd, ready-to-deploy complex
model or set of models that are tuned to optimally perform in the
Pivotal UAP environment.
This option is appropriate for companies with a known business
challenge that could benefit from the injection of analytical
knowledge and capability to address a particularly tough analtyics
or modeling challenge. Examples of possible business challenges
addressable in a 12-week timeframe include: churn drivers,
behavioral targeting, risk models, fraud detection, media mix, and
campaign attribution modeling.

Time is of the Essence


Pivotal UAP brings a platform rich with advanced analytical
capabilities to your data science teams. Capitalizing on that
capability depends on execution of an analytics plan and strategy
that takes time time you may not have. With Pivotal Data
Science Labs, project start-up time shrinks from months to
weeks, accelerated by the efforts of experienced data scientists,
working on your behalf, at your site. The standard packages
of assistance previously described are flexible guidelines, and
customized engagements are encouraged.

Learn More
To learn more about our products, services and solutions, visit us
at goPivotal.com.

Key Benefits of PIVOTAL DATA SCIENCE


LABS

Insights into critical business questions

Training for key analysts

Conversion of existing models

A framework for on-going analytics


- Sample models
- Tools, methodologies, best practices
- Analytics Roadmap

An opportunity to guide Pivotal analytics development

At Pivotal our mission is to enable customers to build a new class of applications, leveraging big and fast data, and do all of this with the power of cloud independence.
Uniting selected technology, people and programs from EMC and VMware, the following products and services are now part of Pivotal: Greenplum, Cloud Foundry, Spring,
GemFire and other products from the VMware vFabric Suite, Cetas and Pivotal Labs.
Pivotal 1900 S Norfolk Street San Mateo CA 94403 goPivotal.com
GoPivotal, Pivotal, and the Pivotal logo are registered trademarks or trademarks of GoPivotal, Inc. in the United States and other countries. All other trademarks used herein are the property of their respective owners.
Copyright 2013 Go Pivotal, Inc. All rights reserved. Published in the USA. PVTL-DS-118-04/13

Vous aimerez peut-être aussi