Transcription
How Do I Move From SAS EnterpriseMiner to SAS Viya?Ask the ExpertMelodie RushGlobal Customer Success Principal Data ScientistConnect with me:LinkedIn: https://www.linkedin.com/in/melodierushTwitter: @Melodie RushCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Melodie RushPrincipal Systems Engineer, SASMelodie is a Principal Data Scientist for the CustomerSuccess Technical Team at SAS Institute. Melodie receivedboth her B.S. in Statistics and her Masters in Science ofManagement from North Carolina State University. Sincejoining SAS, Melodie has developed presentations andmethodology for doing many types of analysis, includingdata mining, machine learning, forecasting, data explorationand visualization, quality control and marketing. She hasspent more than 20 years helping companies identify andsolve problems in each of these analytical areas.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
AGENDAWhat is and WhySAS 9.4 Enterprise Miner (EM) andSAS Viya Visual Data Mining and Machine Learning(VDMML)ComparingEM versus VDMMLTransitioningHow do they work together?Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
What is and Why?SAS 9.4 Enterprise Miner andSAS Viya Visual Data Mining and Machine LearningCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
A Very, Very Brief History of SAS Data Mining/Machine Learning ProductsSAS Institute began in 1976. It’s been a leader in the analytics market ever sinceand its software is used to solve all kinds of analytical problems.SAS 9 was the 9th major update of SAS and is a suite of software for a wide variety oftasks including data management, statistics, and operations research.SAS Viya is the latest update of SAS software. It leverages the latest technologies toenable the powerful analytical techniques SAS is known for.1976SAS Institutefounded1998SAS EnterpriseMiner released2002SAS 9releasedC opyr ig ht SAS I nstitute I nc. All r ig hts r eser ved.2016SAS Viyareleased2017SAS Visual Data Mining& Machine Learningreleased
SAS 9.4SAS Enterprise MinerCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
What is SAS Enterprise Miner ? SAS Enterprise Miner is a graphical userinterface, designed with the specific needs ofdata miners. SAS Enterprise Miner is a data miner’sworkbench that manages the process andprovides a comprehensive set of tools to aidthe data miner throughout the essential steps,known by the acronym, SEMMA: Sample,Explore, Modify, Model, Assess. SAS Enterprise Miner streamlines the datamining process to create highly accuratepredictive and descriptive models based onanalysis of vast amounts of data from acrossan enterprise.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS Enterprise MinerMoving from SAS 9 to SAS ViyaSAS Enterprise Miner on SAS 9SAS VDMML 8.x on SAS ViyaSAS Visual Data Mining and Machine Learning (SAS VDMML) is the equivalent ofSAS Enterprise Miner in SAS Viya. VDMML builds models in the same mannerfollowing the same pipeline format of SAS Enterprise Miner.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS ViyaVisual Statistics and Visual Data Mining and Machine LearningCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
What is SAS Viya?SAS Viyais a cloud-enabled, in-memory analytics engine that providesquick, accurate and reliable analytical insights.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS Viya AnalyticsWhat does it include?Requires Visual AnalyticsVisual Analytics GUISAS Studio UIPython, Java, Lua, R,REST APIsRequires Visual StatisticsVisualAnalyticsVisual StatisticsVisual Data Miningand MachineLearningBaselineProceduresVS ProceduresVDMMLProceduresBaselineAction setsCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.VS ActionsetsVDMMLAction sets
SAS Viya AnalyticsCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Build ModelsSAS Model Studio allows you to build forecasting, text analytics, and machine learning (ML) modelpipelines. You can do everything from data preprocessing to building multiple models to finding your bestmodel all within one pipeline.Main FeaturesAllows for building as manymodels as you wantEmbed open source code withinan analysis and call open sourceas wellCollaborate with others by savingmultiple templates and leveragingthe same environmentUtilize autotuning and AutoMLcapabilities to make the ML modelbuilding process fasterCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Explore and Visualize DataSAS Visual Analytics is the point-and-click visualization tool within SAS Viya and acts as the base visual layer ofSAS Viya. Here, you can build interactive dashboards for reporting with added analytical elements.Main FeaturesPerform ad-hoc exploration of data Discover relationships, trends andoutliersLeverage smart auto-charting andsmart auto- analyticsAnalyze a report and identify issueswith performance and accessibilityCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Develop SAS CodeSAS Studio is the programming interface within SAS Viya. You can program in SAS or CASL (CASL is SAS Viya’sprogramming language) to access capabilities including data quality, machine learning, optimization, textanalytics, forecasting etc.Main FeaturesIncludes autocomplete, syntax help,keyword shortcuts and more forefficient programmingQuickly add prewritten, commonlyused SAS code into your programCreate process flow diagramsAccess basic Git features like cloningrepositoriesCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Why is my Organization Moving to SAS Viya?- Trends Internet of Things Machine Learning and Artificial Intelligence Computer Vision- PowerfulTechnology- Organization’sSkillSet Cloud Containers Graphics Processing Units (GPUs) Business analysts Data scientists ExecutivesCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
What About SAS 9?As your company is making the transition to SAS Viya, there is a good chance youwill still have your SAS 9 environment for some time. SAS 9 will still havemaintenance releases.SAS expects to see customers stand up SAS Viya to meet their new use cases inaddition to continuing to use SAS 9 for their existing deployments.The reason for this is because it will take some time before full replacementvalue is realized in SAS Viya. As well, this transition period takes intoconsideration the process of adopting this new technology for existing SAS users.To that end, SAS has ensured that these two environments can betechnologically bridged.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS 9 and SAS Viya: Main Similarities and DifferencesThough there are many similarities and differences between the two solutions, here arethe main ones to consider:What’s the Same? The ability to write code or do work ina drag-and-drop interface The ability to leverage SAS’intellectual property (IP) to accelerateyour analytical processesWhat’s Different? The underlying architecture SAS Viya products are located centrallyand accessed in one central URL whereasSAS 9 has multiple user interfaces toaccess different products Products in SAS Viya have substantialfeature improvements. As well, SAS Viyahas a modern UI for better end-userexperienceCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
CompareSAS Enterprise Miner and SAS Visual Data Mining andMachine LearningCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
EM and VDMMLThe difficulty in comparing1.2.3.4.EM only has one interface, VDMML has three ways to interact.Some functionality has moved to other packages.VDMML is still growing.Model Studio isn’t a product.EMVDMMLCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
EM and VDMMLAnalytical Algorithm DifferencesA (EM)Incremental ResponseTime Series Nodes(6)†Dmine RegressionRule InductionRules BuilderTwo StageMBRASOMB (VDMML)ABBCommon elements (AB)listed in Appendix† All ETS-like functionality for Viya was moved to VFAnomaly DetectionSVDDRPCA*MWPCADeep mension d ModelsRandom ForestGaussian MixedMulti-task LearningQuantile RegressionModel-based ClusteringSemi-supervised LearningNetwork AnalysisSentiment Analysis* Also used for Dimension ReductionCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
EM and VDMMLFeature DifferencesEMVDMML REST API callsAuto-tuningThe ExchangePipeline templatesAutomatic pipelinegeneration Model interpretability Automatically generatednatural language toexplain results (NLG) Deep Learning (DLPy)Target profilerDecisions nodeReporter nodeControl point nodeGroup-by processing*Custom Extension NodesRapid Predictive Modeler EMVDMMLCommon elements are listed in AppendixCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
EM and VDMMLExecution Differences*VDMMLEnterprise MinerOnly 1 GUI (Java based)Only 1 level of execution (PROC)Supports multiple data sourcesRuns SAS MacrosTasksSAS PROCSCAS PROCSCASL* Most of these are differences between 9.4 and Viya not necessarily EM and VDMML.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.3 GUIs (HTML5 based)Visual (“VA”)Model StudioSAS StudioAPI InterfacesPythonRJava3 levels of executionTask levelPROC levelAction levelSupports only single datasource
EM and VDMMLArchitectural Differences*Enterprise Miner Mostly single-threadedSymmetric Multi-Processing (SMP)Supports MVADesktop client optionVDMMLViya ServerEMServer9.4*These are differences between 9.4 and Viya (not including HPA)Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved. Mostly multi-threadedMassively Parallel ProcessingDistributed ComputingSupports concurrencySupports multi-tenancyEasily deployed in the cloudLINUX only
EMSAS 9.4 and SAS Viya Functional Comparison Technical PaperEMVDMMLCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.VDMML
TransitioningMoving from SAS 9 to ViyaCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
ProcessOrganization of the toolsModelsIncluding EM models in PipelinesIncluding Viya Models in EM FlowsScoringUsing EM score code in ViyaUsing Viya score code in SAS 9.4Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
ProcessWhat happened to SEMMA?Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
List of SAS Enterprise Miner sionTreeMODELDataPartitionAppendFile valCredit Scoring*CorrelationAnalysisHP ExploreHP BayesianNetworkHP End GroupsStart GroupsHP TransformHP ImputeModelComparisonOpen tial LeastSquaresLink AnalysisMarket MDBHP VariableSelectionRuleInductionEnsembleTS DimensionReductionHP ForestHPDecisionTreeScoreSegment ProfileReporterScore CodeExportCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.GradientBoostingTSDecomp.HP lderReplacementHP NeuralVariableSelectionPath AnalysisDM NeuralTS DataPrepInputDataSampleMergeFilterLARSTwo StageMBRModel ImportTSSimilarityHP GLMHP SVMTS ExponentialSmoothingHPClusterSAS CodeSaveRegisterMetadata*Requires Credit Scoring for SAS EnterpriseMiner Add-on LicenseDataExt DemoMetadataHP PrincipalComponentsSASViyaCode
SAS Enterprise Miner A SEMMA ProcessSAMPLEEXPLOREMODIFYCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.MODELASSESS
SAS Visual Data Mining and Machine LearningPipelinesCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS Visual Data Mining and Machine LearningPipelinesDataPreprocessingSupervised LearningModel ComparisonCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
ModelsIncluding EM Models into PipelinesCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Including Existing EM models into PipelinesEM ProcCodeEM ProcCode(ASTORE)EM ScoreCodeEMASTORECodeCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.EM BatchCodeViyaModel
EMScoreCodeData StepCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesSAS Code Score Code Create Enterprise Miner Diagram Include Score Node and Score Code Export NodeCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesSAS Code Score Code Select Output Directory for the Score Code ExportCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesSAS Code Score Code Select Output Directory for the Score Code ExportCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesSAS Code Score Codescore.sasCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesSAS Code Score Code In Pipelines add a score code import node (under SupervisedLearning) Open Code EditorCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesSAS Code Score Code Select DATA step code Browse and select local codeCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesSAS Code Score CodeCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
EMScoreCodeASTORECopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE) What is an ASTORE file?– An Analytic Store, or ASTORE, is a system that allows the state of a trainedpredictive model to be saved in a transportable form.– This enables it to subsequently be used to score new data in a variety ofenvironments.– A key feature of an ASTORE is that it can be easily transported from oneplatform to another. What models support ASTORE?– Many SAS analytical procedures save the results from the training phase ofmodel development as ASTORE models. Nodes in EM that create ASTORE filesHP SVM and HP Forest.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE) Create Enterprise Miner Diagram Include Score Node and Score Code Export NodeCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE) Select Output Directory for the Score Code ExportCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE)epscore.sas includes score code and link to ASTORE file(ep entire pipeline or process)Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE) In Pipelines add a SAS code node (under Miscellaneous) Open Code EditorCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE) In Pipelines add a SAS code node (under Miscellaneous) Open Code Editor – add code to load ASTORE file into memoryCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE)Note: that the analytic store does not reside on your client but must reside in apath available to the SAS client (the SAS Viya session).Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE) In Pipelines add a score code import node (under SupervisedLearning) Open Code EditorCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE) Select Analytic store code Browse and select local code (epscore.sas) Browse and select ASTORE table loaded in MemoryCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Score Code to Viya PipelinesAnalytic Store Code (ASTORE)Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
EMPROCCodeSAS ProceduresCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc DMREG In Pipelines add a SAS code node (under Miscellaneous) Right mouse click and move to Supervised Learning Open Code EditorCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc DMREG code – part 1EM procedures require data to be split into training and validation data setsCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc DMREG code – part 2EM procedures require Data Mining Database (dmdb)Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc DMREG code – part 3Proc DMREG codeCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc DMREG code – part 4Display plot of parameter estimatesCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc DMREG code – ResultsDisplay plot of parameter estimatesCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc HPSVM (create ASTORE file) In Pipelines add a SAS code node (under Miscellaneous) Right mouse click and move to Supervised Learning Open Code EditorCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc HPSVMData already split into Training and Validation data sets and DMDB created inthe DMREG code in previous exampleCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Procedure Code to Viya PipelinesProc HPSVM code – ResultsCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
EMBatchCodeProcess Flow CodeCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Batch Code to Viya PipelinesBatch Code Node Create Batch Code from Enterprise Miner Diagram Right Mouse Click on the Score Node & Select ExportPath as SAS ProgramTip: Save the code where it can be uploaded via your browser to SAS Viya.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Batch Code to Viya PipelinesBatch Code NodeTip: Save the code where it can be uploaded via your browser to SAS Viya.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Batch Code to Viya PipelinesBatch Code Node In Pipelines add a Batch Code Node (under Supervised Learning)Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Batch Code to Viya PipelinesBatch Code Node Open Code EditorCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Batch Code to Viya PipelinesBatch Code Node Select the batch file saved from Enterprise MinerCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding Enterprise Miner Batch Code to Viya PipelinesBatch Code Node Now the Enterprise Miner Diagram winner is part of your pipelineCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Simplified SAS Viya ArchitectureSAS Viya ProductsVisualAnalyticsVisualStatisticsVisual Data Miningand Machine LearningController NodeVisual TextAnalyticsVisualForecastingMicroservices-based mid-tierSAS Viya EnginesCAS WorkerNode 1CAS WorkerNode 2 CAS WorkerNode nCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.SAS ComputeServer(minimal functionality,for Viya-onlydeployments)
Simplified SAS Viya ArchitectureUsing Enterprise Miner Batch Code: ConsiderationsMicroservicesbased mid-tierControllerNodeThe Enterprise Miner batch code doesnot execute in CAS.SAS Viya EnginesCAS WorkerNode 1CAS WorkerNode 2 CAS WorkerNode nSAS ComputeServerIt executes on the SAS Compute Server.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS Visual Data Mining and Machine LearningUsing Enterprise Miner Batch Code: Considerations When invoked from a Model Studio pipeline in Viya, the batch codefrom SAS Enterprise Miner does not execute in CAS, rather in theSAS Viya Workspace Server.– A sample of the input table loaded into CAS is transferred to the SAS ViyaWorkspace Server.– After the batch code runs, Model Studio retrieves either the model score codeor the analytic store and EPCODE.– The retrieved code is used to score the original input data in Model Studio. Thisscored data is used to produce the assessment results. Typically, the server with the SAS Viya Workspace Server is within theViya cluster, so the transfer of the data is limited only by theconnection between the nodes.DocumentationCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Model ComparisonCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Model ComparisonCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
ScoringIn ViyaCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Scoring in SAS Viya Score Code Import Node in Pipeline (Supervised Learning) Score Data Node in Pipelines (Miscellaneous) Create Score Code or ASTORE file from– modeling nodes– champion models SAS Studio Scoring Task (SAS Viya Evaluate and Implement Models)Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Scoring in SAS ViyaScore Data Node in PipelinesCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Scoring in SAS ViyaDownload Score Code: Any Modeling Node To save the score code of a modeling node,right-click the node and select DownloadScore Code. A Zip file will be saved in your browser’sDownloads folder. Within the Zip file will be a SAS programcalled dmcas epscorecode.sas thatcontains the EP score code.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Scoring in SAS ViyaDownload Score Code: Champion Model In the Pipeline Comparison, click the project pipeline menu (snowman) inthe upper rightSelect Download Score Code. (It will download the code for the championmodel, which in this example, is the Ensemble.)Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Score Code in SAS Viya Some model types generate score code as an ASTORE. An additional file will alsoneed to be downloaded. A comment in the top of the downloaded score code will indicate which ASTORE(s)are referenced and in which CASLIB on the SAS Viya server they are stored. The default location is a CASLIB called Models that by default is located Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Score Code TypesNode NameAnomaly DetectionBayesian NetworkClusteringDecision TreeEnsembleType of Score CodeAnalytic storeAnalytic storeDATA stepDATA stepDATA step (if all models produce DATA step), otherwise one or moreanalytic stores and the EP score code to combine the models’ score codeFeature ExtractionFilteringForestGLMGradient BoostingImputationLinear RegressionLogistic RegressionNeural NetworkDATA stepDATA stepAnalytic storeDATA stepAnalytic storeDATA stepDATA stepDATA stepDATA step for networks less than 6 layers, analytic store for networks with6 or more layersReplacementSVMText MiningTransformationsDATA stepAnalytic storeAnalytic storeDATA stepCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.This table demonstrateswhich Model Studionodes produce scorecode, as well as the typesof code they produce.Documentation link
Scoring in SAS ViyaScoring Task in SAS StudioCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
ModelsIncluding VDMML models in EM diagramsCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Including SAS Viya Nodes into EM Diagrams Viya Code Node Scoring Viya Models in SAS 9.4Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding SAS Viya Nodes to SAS Enterprise MinerVDMML Models or Viya Procedures Allows you to run VDMML Models in CAS within EM diagrams in orderto compare modelsCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Adding SAS Viya Nodes to SAS Enterprise MinerViya Code Node Under the Utility Tab SAS Viya Code node is available in the Utility nodes.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS Enterprise Miner Viya Code Node Just like in the other code nodes (SAS Code, Open Source Integration), severalmacros and macro variables have already been defined for you and some codehas already been included to streamline the process for you. In invokes macros to: Streamline the pre-modeling process:– Signing into the Viya server– Loading your data into CAS– Setting up macros and macro variables for your data (target, inputs, rejected, etc.) Streamline the post-modeling process:– Model assessment– Model selection– Report generation– Viya session terminationCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS Enterprise Miner Considerations In order to run Viya algorithms from Enterprise Miner, the input datato the Viya node needs to be transferred to the Viya environment. Typically, this isn’t a problem when the environments are co-located inthe same data center, or data centers with secure, fast connectionsbetween them. However, when the data is large or the environments are not colocated, the movement of data may not be practical, feasible orallowed (by your corporate network/security policies) depending onyour overall architecture.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
ScoringViya Models In SAS 9.4Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Scoring SAS Viya Models in SAS 9.4 Data Step code can be run directly in SAS 9.4 without any changes(except to add the data step datasets). Using SAS Viya Data Step Codein SAS 9.4 Analytic store (ASTORE) will need to have the ASTORE filesdownloaded to your SAS 9.4 environment and use the EP (entirepipeline) score code file. Using SAS Viya ASTORE Score CODE in SAS9.4Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Scoring SAS Viya Models in SAS 9Data step score code exampleNeed to insert aData Step to identifyincoming andoutgoing data.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Scoring SAS Viya Models in SAS 9ASTORE and EP Score code exampleEP Score code file name is dmcas epscorecode.sasCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
ResourcesWhere to learn moreCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
ResourcesWhere to learn more SAS Enterprise Miner and SAS Visual Data Mining and MachineLearning Handshake Make SAS Enterprise Miner Play Nicely with SAS Viya SAS9.4 and SAS Viya Functional Comparison Technical Paper SAS Viya Overview SAS Enterprise Miner Support Page SAS Visual Data Mining and Machine Support Page Create an ASTORE Using SAS Enterprise MinerCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
SAS Enterprise Miner Resources for Viya Code Node Make SAS Enterprise Miner Play Nicely with SAS Viya -global-forumproceedings/2018/2204-2018.pdf Github examples: https://github.com/sassoftware/em-bridge2viya YouTube Video: Use SAS Enterprise Miner to Run Machine LearningAlgorithms in SAS Viya Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
CommunitiesCopy rig ht SA S Institute Inc. A ll rig hts re se rve er.sas.comCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Questions?Thank you for your time and attention!Connect with me:LinkedIn: https://www.linkedin.com/in/melodierushTwitter: @Melodie Rushsas.comCopyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Explore Helpful User ResourcesAsk the ExpertView other user webinars that provide insights into using SAS products to make your job easier.FREE TrainingLearn from home – free for 30 days. Get software labs to practice and online support if needed.SAS Support CommunitiesAsk questions, get answers and share insights with SAS users.SAS Analytics ExplorersAn exclusive platform to collaborate, learn and share your expertise. Gain access to a diverse network to advance your career. Specialrewards and recognition exclusively for SAS users.SAS Users YouTube ChannelA plethora of videos on hundreds of topics, just for SAS users.NewslettersGet the latest SAS news plus tips, tricks and more.Users GroupsMeet local SAS users, network and exchange ideas – virtually.SAS ProfileIf you haven’t already done so, create your SAS Profile to access free training, SAS Support Communities, technical support, softwaredownloads, newsletters and more.Copyr i ght SAS I nsti tute I nc . Al l r i ghts reser ved.
Thank youfor joining us forthis SAS webinarC opyr ig ht SAS I nstitute I nc. All r ig hts r eser ved.
SAS Enterprise Miner is a graphical user interface, designed with the specific needs of data miners. SAS Enterprise Miner is a data miner's workbench that manages the process and provides a comprehensive set of tools to aid the data miner throughout the essential steps, known by the acronym, SEMMA: Sample, Explore, Modify, Model, Assess.
