klog pÅ sas · sas data integration studio you can use sas data integration studio on sas 9.4 to...
TRANSCRIPT
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
KLOG PÅ SAS
VIYA UPDATE
TECHNICAL ACCOUNT MANAGER
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
executable anywhere data is (on-disk, in-
stream, in-database, in-memory, in-device)
SAS will analyze data of any size
in any form (table, stream, web service)
in any location (on-premises, cloud, hosted, hybrid)
from any source (private, public, partner)
by any language (SAS, Python, Java, REST-API)
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
Web
browser
Other
client
Smart
device
RDBMS Hadoop
RESTful APIsMicroservices In-Memory Engine
SAS Studio &
Python/Java/Lua
interfaces
SAS
Environment
Manager
Data Connectors / Accelerators
In-Database / embedded processes
VIYA PLATFORM
Cloud Analytic Services (CAS)
Work-
space
server
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
Visual
Analytics
Visual
Statistics
Visual Data
Mining and
Machine
Learning
Visual Data Mining and Machine Learning
Baseline
Procedures
VS
Procedures
VDMML
Procedures
Visual Analytics GUI
VS
Action sets
VDMML
Action setsBaseline
Action sets
Python API
SAS Studio UI
…..
…..
…..
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
April Summer Fall
SAS®
Visual Data Mining
and Machine Learning
SAS®
Visual Analytics
SAS®
Visual Statistics
SAS®
Visual Investigator
2017
SAS®
Studio
SAS®
Data Management
• Access
• Data Quality
• Data Prep
SAS®
Visual Scenario
Designer
SAS®
Analytics Suite
SAS®
Analytics as a
Service
etc….
Early Adopters SAS®
Visual Data Mining
and Machine Learning
SAS®
Visual Investigator
Early Adopters
SAS®
Visual Scenario
Designer
SAS® Viya™ announced
Early Preview
SAS®
9.4 M4
SAS®
Visual Analytics
SAS®
Visual Statistics
SUBJECT TO CHANGE
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
2016
2017
SUBJECT TO CHANGE
VDMML, VA, VS Visual Investigator
VDMML, VA, VS VDMML, VA, VS
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
SAS VIYA - SAS STUDIO
DEMO
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
SAS®
9
• SAS®
9 data accessible within SAS®
Viya™
• Remotely execute code from / to both
• SAS®
Viya™
models publishable to SAS®
9
Bridging SAS® 9 and SAS® Viya™
Data Code DataModel ModelModelCodeData
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
SAS 9.4
2016 2017 2018 …
SAS Viya
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
USECASE
Hvad kan VDMML bruges til – nu ?Innovation Lab
Machine Learning
Data Discovery / Explorering
Data Science
Advanced Analytics
SAS Viya
Big Data
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
SAS®Viya™ Programming
REST APIs
SAS® 9
Next-gen Solutions
Visual Statistics
Visual Analytics
Visual Investigator
Visual Data Mining
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
Open platform with a common code base that is…• Executable anywhere (in-Memory, in-Database, in-
Hadoop, in-Cloud, in-Stream, in-Device)
• Portable to any IT environment (desktop, server, grid,
cluster, or cloud)
• Accessible from third party applications
• Available in any public / private cloud
• Capable of management and inventory of all analytics
assets
• Infused with sophisticated and native search that does
not require pre-defined schemas in applications
Support for the Analytics Lifecycle• Data Management
• Visual Data Exploration
• Interactive Discovery
• Advanced Analytics
• Model Versioning & Inventory
• Decision Management
• Reporting
• Dashboards
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
SPØRGSMÅL?
TECHNICAL ACCOUNT MANAGER
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
Following are several examples of SAS 9.4 solutions that contain user interface features that you can use to issue SAS/CONNECT
statements to bridge the SAS 9.4 and SAS Viya environments:
SAS Enterprise MinerFrom your SAS Enterprise Miner process flow diagram running on SAS 9.4, you can remotely submit SAS Data Mining and Machine Learning
procedures to run in SAS Viya and then return the results to SAS 9.4 and integrate them into the SAS Enterprise Miner output.
SAS Model ManagerYou can run SAS Data Mining and Machine Learning procedures and train models in your SAS Viya environment, and then upload the models
and the models’ metadata to SAS 9.4. From SAS 9.4, you can remote submit code to register the SAS Viya created model and metadata in SAS
Model Manager. You can use SAS Model Manager to manage both your SAS 9.4 models and your SAS Viya models. For example, you can use
SAS Model Manager to deploy score code derived from both SAS 9.4 and SAS Viya models to a database.
SAS Data Integration StudioYou can use SAS Data Integration Studio on SAS 9.4 to perform tasks such as fetching data from multiple sources and preparing the data for
mining. Then, you can use the Data Transfer node to upload the data table to SAS Viya and load it directly into memory.
SAS Enterprise Guide / SAS Add-In for Microsoft OfficeYou can use the custom code nodes in either of these products to upload or download data between the two environments. For example, you
could remote submit code to SAS Viya for processing, and then return the computational results back to SAS Enterprise Guide on SAS 9.4.
SAS StudioYou can use the SAS Studio web interface to perform many programming actions in SAS Viya. For example, you could enter code in SAS Studio
to access data in your SAS Viya environment, remotely submit SAS Viya procedures to run on the data, and then transfer the results data
between the two environments.
SAS Stored ProcessesYou can save SAS/CONNECT SIGNON, RSUBMIT, UPLOAD, and DOWNLOAD statements as SAS Stored Processes,
and then use the SAS Stored Processes as another way to bridge your SAS 9.4 and SAS Viya environments.
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
SAS Viya Does Not Replace SAS 9
SAS 9 and SAS Viya are two unique platforms built for different use cases. However, these platforms share
some functionality. For example, SAS 9 uses the SAS programming language, and SAS Viya uses the next
generation of SAS programming with the new CAS programming language. The CAS language is very similar
to the SAS language. Some procedures are available on both platforms, so some existing SAS code can be
run on the SAS Viya platform. However, SAS Viya also contains new procedures that take advantage of the
open, distributed environment. As a result, some SAS 9 procedures do not exist in the new SAS Viya
environment.
• Because of the advancement in technology and infrastructure since SAS 9 was released, the technology
of SAS Viya is very different from SAS 9. Here are some examples: In SAS Viya, SAS Cloud Analytics
Services is the third-generation of in-memory services from SAS. It replaces SAS LASR Analytic Server
and the SAS High-Performance Analytics engine in SAS 9.
• For functionality other than CAS, there is no concept of roles or capabilities in SAS Viya. Access to
functionality is managed by rules that target a service, a service endpoint, a media type, or a pseudo URL.
• Instead of the metadata-based authorization, SAS Viya maintains authorization within the CAS engine and
by using host identities.
• In SAS Viya, images are a recognized data type.
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
SAS®
Viya™
…WITH SAS HIGH PERFORMANCE ANALYTICS
• SAS High Performance Data Mining
• SAS High Performance Statistics
• SAS High Performance Text Mining
• SAS Visual Data Mining and
Machine Learning
• SAS Visual Statistics
What it is In-memory analytic processing
algorithms/methods specifically designed for big
data, complex analysis and fast results from
built-in workflow.
In-memory analytic processing
algorithms/methods optimized for
parallel processing and initially
programmatically available. Designed
for big data, complex analysis, fast
results and open to be used from 3rd
party languages
Who it’s for Dir / VP IT, BI; LOB Dir / VP/ Lead; CAO, VP /
Dir Analytics; Data Scientist / Miner/ Statistician
Data Scientist / Miner / Statistician,
Analytic and Application Engineers
Copyr i g ht © 2016, SAS Ins t i tu t e Inc . A l l r ights reser ve d .
SAS®
Viya™
…WITH ENTERPRISE MINER
• SAS Enterprise Miner 14.1 • SAS Visual Data Mining and Machine Learning 8.1
What It Is • An integrated drag-and-drop process flow and programming
interface streamlining the end to end data mining and
machine learning process.
• SAS’ complete library of state-of-the-art statistics, data
mining, machine learning, and text mining capabilities. New
methods from VDMML accessible via the SAS 9.4 bridge.
• Enables automated batch processing and model retraining
• One, scalable in-memory processing environment.
Integration with SAS Grid Manager. Windows, Linux, AIX,
etc…
• Scorecode deployable via SAS, Java, C, in-database or via
web-services in SAME evironment
• Seamlessly integrates with SAS Model Manager 9.4 to
completely enable the analytics life cycle.
• Can call on open source models from within EM. Generates
PMML.
• Supports data mining and machine learning process via two independent
interactive and programming interfaces.
• Growing library of statistical, data mining, machine learning, and text mining
capabilities. Integrated optimization into methods using autotuning.
Basic deep learning capabilities.
• Library for reading in image files directly into SAS environment
• One, scalable in-memory processing environment. Failover, elastic, cloud-
friendly. Linux-only.
• Scorecode deployable via SAS, Java, C, in-database or via web-services
USING THE BRIDGE TO 9.4
• Integrates with SAS Model Manager 9.4 USING THE BRIDGE TO 9.4
• Can call actions through APIs for Java, Python, Lua and REST API
Who It’s ForData scientists, statisticians, data miners, engineers,
researchers and scientists
Data scientists, statisticians, data miners, engineers, researchers and scientists.
New and existing customers who are interested in leveraging the latest
machine learning algorithms and don’t mind the tradeoffs.