WIXLIB

SAS Visual Data Mining and Machine LearningIntroductionCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Data Mining and Machine Learning Visual ”drag & nVisualExplorationModelDeploymentCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.MachineLearningModelStudio

Data Preparation Access to different datasources Training-Validation DataPartitioning Feature Engineering (e.g.parameters, interactions) Variable selection andmissing valuesCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Visual Exploration Interactively discoverrelationships, trends, outliers Smart autocharting Analytics driven visualizations Explore predicted outputs Variable transformationCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Model Studio Pipeline of activitiesDrag and drop and access tocodeNodes are run asynchronouslyReproducibilitySAS best practice toolkitCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Modern Machine Learning Forest Neural Network (including DeepLearning) Gradient Boosting Support Vector Machines Factorization Machines Bayesian Networks AutotuningCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Comparison and DeployModel comparisonsummaries Interactively assess models Assessment charts forpartitioned data Publish score code; batch,API call, in-database Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Data Mining and Machine Learning 8.3 Visual InterfaceMachine Learning Techniques Forest Factorization Machine Gradient Boosting Neural Network Support Vector MachineCommon Features Training-Validation Model Assessment Model Comparison Score Code or Astore Table Ability to export model statistics into ExcelCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Data Mining and Machine Learning SAS Studio – Available Programming TasksCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Data Mining and Machine Learning Capabilities via Programming InterfacesAnalyticalAnalytical FOREST (Random Forest Model)GRADBOOST (Gradient Boosting)NNET (Neural Network)SVMACHINE (Support Vector Machine)FACTMAC (Factorization Machine)NETWORK (Network analytics and communitydetection)TEXTMINE (NLP and statistical analysis of text)TMSCORE (Scores textural data)BOOLRULE (Boolean rules extraction)MTLEARN (Multi Learning task for least squaresloss)SEMISUPLEARN (graph-based semisupervisedlearning algorithm) SVDD (Support Vector Data Description)MWPCA (Moving Window Principal ComponentAnalysis)RPCA (Robust Principal Component Analysis)TSNE (t-distributed stochastic neighborembedding)ASTORE (Analytic Store for Models)FISM (Frequent Item Set)MBANALYSIS (Market Basket Analysis)GVARCLUS (Graphical Variable Clustering)FASTKNN (K- Nearest Neighbor)BNET (Bayesian Network)Deep Learning (no procedures, through CASactions)Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Data Mining and Machine LearningAction SetsAnalytic Store ScoringAssociation Rule MiningAudio ActionAutotuneBayesian Net ClassifierBioMedImageBoolean RuleFactorization MachineFast k-Nearest Neighbors AlgorithmGeneralized Linear Multitask LearningGraph-Based Semisupervised LearningGraphical Variable ClusteringImageLanguage ModelMLToolsNetworkNeural NetworkNonparametric BayesRobust PCASupport Vector Data DescriptionSupport Vector MachineText MiningTSNELink to documentationCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Data Mining and Machine LearningVisual Interface DemoCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Text AnalyticsCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Unearthing the full potential within complex data sources canbe trickyLanguage is messy! Large data volumes and inconsistent formats Multiple sources and languages Misspellings, slang, and abbreviations Highly subjective to interpretation and contextManual review is both inconsistent and time consuming,and a sampling approach can mean missing out onvaluable information and the big pictureCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS is continually extending its Natural Language Processing capabilitiesNatural LanguageProcessing (NLP)A branch of artificial intelligence that helps computersunderstand, interpret and manipulate human language.Parsing and Entity/RelationshipExtractionTopic Detection,Text Clustering & ProfilingNatural LanguageUnderstanding & GenerationNLPClassification (Categories, Sentiment)Automatic Summarization,SearchAuto-Translation* & Speech toTextCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.* An area of NLP not currently offered by SAS

SAS Natural language processing at a glanceSAS Text Analytics solutions perform document conversion*, tokenization, lemmatization ( stemming), part-ofspeech detection, and apply lexicons for misspellings*, synonyms, multi-word terms, and start/stop wordsImport DataI disputed a crediter for the incorrect amt they reported toTrans Union credit bureau.Parse andDetect EntitiesI disputed a crediter for the incorrect amt they reported toTrans Union credit bureau.Filter, Tag, andResolve TermsI disputed a crediter for the incorrect amt they reported toTrans Union credit bureau.Automatically detect termvariants, misspellings andparts-of-speechCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.Resolve shorthand, slang,and abbreviations withsynonym lists

Global Language ugueseCopy rig ht SA S Institute Inc. A ll rig hts re se rve gThaiTurkishVietnamese

SAS Visual Text Analytics Visual InterfaceText Analytics Techniques Contextual Extraction Text Parsing Sentiment Analysis Develop Topics Content CategorizationCommon Features Report ready output Model ready output ScoringCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Visual Pipeline Customizable and portable nodes andpipelines Quick save of output data from Topics andCategories Easy access to automatically-generated scorecode Native integration and common interface withdata preparation, visualization, modelmanagement, data lineage, and otheranalytics solutionsCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Inspect and Manage Terms View Kept/Droppedterms and theirvariants, roles, andfrequency in yourdocument collection Explore Term Mapsand Similar Terms Review matcheddocuments to seeterms used in contextCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Explore and Manage Discovered Topics Text Topics arediscovered automatically,without requiring explicitinput from the user View the terms anddocument comprising atopic and how pervasiveeach topic is in thecollection Merge or split topics,create custom topicsfrom selected terms, orpromote topics tobecome rule-basedcategoriesCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Content Categorization Build, test, and assess categorydefinitions and manage yourtaxonomy structure Leverage machine-built rules, orauthor your own definitionsusing keywords and a broadrange of Boolean and linguisticoperators & qualifiers Use Textual Elements and therule-builder to help write rulesfrom scratch Produce report-ready ormodeling-ready categorizedoutputCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Contextual Extraction Leverage pre-defined entitiesfor standard concepts such asdates, locations, andmeasures Create custom entities using abroad range of rule types,operators, regularexpressions, and linguisticqualifiers Extract custom facts(relationships betweenentities)Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Sentiment AnalysisAvailable languages for domain-independent sentiment model Create custom entities using abroad range of rule types,operators, regularexpressions, and linguisticqualifiers Arabic Chinese (Simplified) Italian Chinese (Traditional) Japanese Dutch Korean English Portuguese Extract custom facts(relationships betweenentities) Farsi Spanish French TurkishCopy rig ht SA S Institute Inc. A ll rig hts re se rve d. German

SAS StudioVTA Pre-built tasks Wizard-driven tasks Options to use algorithmsnot presented in ModelStudio GUI Easily access intermediateand output tables Auto-generate SAS codefor further developmentCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Text Analytics Capabilities via Programming InterfacesAnalytical TEXTMINE (NLP and statistical analysis of text)TMSCORE (Scores textural data)BOOLRULE (Boolean rules extraction)Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Text AnalyticsAction SetsBoolean RuleText MiningDeep RNNText ParseLDA Topic ModelingText Rule DevelopmentSearchText Rule DiscoverySearch AnalyticsText Rule ScoreSentiment AnalysisText SummarizationSmart Data Set AnalysisText TopicsTA Conditional Random FieldsText UtilitiesLink to documentationCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Text AnalyticsOpen SourceCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Consumer ComplaintsOur example today The dataset is from the Consumer Financial Protection Bureau (CFPB) andcontains complaints submitted by customersThe goal of this exercise is to analyze the text data from their complaintsThe data are at the customer-level (subject-level)n 2.4Kcolumns 18Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Consumer ComplaintsDetailsCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Consumer ComplaintsSample DataCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Consumer ComplaintsProfileCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

NarrativesConsumer Complaints Submitted to CFPBCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

CategorizationDocuments are categorized as related to Products, Overdrafts, Legal Issues orDigital Currency.Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SentimentDo people feel positive, negative or neutral toward something?Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Concept ExtractionExtract specifics, like interest rate, from the text.Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Topic DiscoveryWhich topics appear in my document collection?Copy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Visual Text AnalyticsDemoCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

End-to-End Analytical CapabilitiesData ManipulationIn-Memory Data StepFrequency / CrosstabData TransposeVariable BinningVariable Cardinality AnalysisVariable SummarySampling and PartitioningMissing Value ImputationVariable SelectionModel AssessmentDS2FedSQLS3 Data ConnectorText AnalyticsWord CloudTopicsText ParsingContent CategorizationSentiment AnalysisContextual ExtractionStatisticsCox Proportional HazardsDecision TreesDesign MatrixGeneral Additive ModelsGeneralized Linear ModelsIndependent Component AnalysisK-means and K-modes ClusteringLinear RegressionLinear Mixed ModelsLogistic RegressionModel-Based ClusteringNonlinear RegressionOrdinary Least Squares RegressionPartial Least Squares RegressionPearson CorrelationPrincipal Component AnalysisQuantile RegressionShewhart Control Chart Analysis For documentation please click here Optimization and Forecasting are alsosupportedCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.Machine LearningBayesian NetworksBoolean RulesFactorization MachinesFrequent Item Set MiningGaussian Mixture ModelGradient BoostingK Nearest NeighborImage Processing (including Biomedical)Market Basket AnalysisMoving Windows PCAMultitask LearningNetwork Analytics / Community DetectionNeural Networks / Deep LearningRandom ForestRobust PCASemi-supervised LearningSupport Vector Data DescriptionSupport Vector Machinest-distributed Stochastic Neighbor EmbeddingText MiningVariable Clustering

ResourcesCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS ViyaResourcesVideos 6 minutes getting started video5 minutes getting started video8 minute demo videoFeature Engineering videoCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS ViyaResourcesSAS Visual Statistics User’s al-statistics/index.html#s1 2SAS Visual Data Mining and Machine Learning User’s al-data-mining-machine-learning/index.html#s1 1SAS Visual Text Analytics User’s al-text-analytics/index.htmlOverview, Training, Samples and Tips SAS Viya Overview SAS Viya Training A Beginner’s Guide to Programming in the SAS Cloud Analytics Services EnvironmentCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

ResourcesProgramming SAS StudioCAS actions documentationSAS Github page for SWAT-PythonSAS Github page for SWAT-RMore example scripts for using SWAT-R & SWAT-PythonCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Useful WebsitesDeveloper.sas.com, Communities.sas.comCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

Questions?Thank you for your time and attention!Connect with me:LinkedIn: https://www.linkedin.com/in/melodierushTwitter: @Melodie Rushsas.comCopy rig ht SA S Institute Inc. A ll rig hts re se rve d.

SAS Viya is an underlying foundation for additional products that will take advantage of a cloud-enabled, open platform. Most offerings include both a coding interface as well a visual interface. SAS Visual Analytics SAS Visual Statistics SAS Visual Data Mining and Machine Learning SAS Visual Forecasting SAS Visual Text Mining