Copyr i g ht 2014, SAS Ins titut e Inc. All rights res er ve d. WHAT S NEW IN SAS ANALYTICS 9.4

Transcription

1 WHAT S NEW IN SAS ANALYTICS 9.4

2 AGENDA SAS ANALYTICS 13.1 SAS Enterprise Guide 6.1 SAS/STAT 13.1 SAS/ETS 13.1 SAS Enterprise Miner

3 SAS ANALYTICS 13.1 LATEST RELEASE OF SAS ANALYTIC SOFTWARE SAS ANALYTICS 13.1 was released in December Available on all platforms Requires Base SAS 9.4M1 Builds on the 12.1 and 12.3 releases See doc/miner/ for further information.

4 WHAT S NEW IN SAS ENTERPRISE GUIDE 6.1

5 SAS ENTERPRISE GUIDE 6.1 Release Objectives Leverage the capabilities of SAS 9.4 platform Ease-of-use for new Enterprise Guide users More for the developer New Analytic tasks

6 SAS ENTERPRISE GUIDE 6.1 WHAT S NEW The ability to analyze a SAS program to determine whether there are any possible internationalization issues. When you analyze a program for internationalization, SAS Enterprise Guide lists the lines of code that might be affected and suggests substitutions when possible Administration enhancements, such as the new stand-alone installer and application streaming support. The new installer is much smaller, thereby making it easier to install over a distributed deployment, especially using provisioning tools such as System Center Configuration Manager (SCCM).

7 SAS ENTERPRISE GUIDE 6.1 WHAT S NEW Improved programmer productivity with the new Log Summary window, which lists all the errors, warnings, and notes that were generated when the program ran, as well as related line numbers and a sample of the affected code The ability to use notes to add information to a process flow or to specific objects in the process flow Task Gallery Integration with SAS high-performance tools with the addition of the High-Performance Logistic and High- Performance Linear Regression tasks Improved Scheduler Windows Scheduler integration

8 SAS ENTERPRISE GUIDE 6.1 FOR THE PROGRAMMER/DEVELOPER Code or Log tab

9 SAS ENTERPRISE GUIDE 6.1 NOTES Note

10 SAS ENTERPRISE GUIDE 6.1 NOTES Note linked to program (or Task)

11 SAS ENTERPRISE GUIDE 6.1 TASK GALLERY

12 SAS ENTERPRISE GUIDE 6.1 TASK GALLERY ZOOM

13 SAS ENTERPRISE GUIDE 6.1 NEW RFM TASK

14 SAS ENTERPRISE GUIDE 6.1 HIGH PERFORMANCE PROCEDURES High performance

15 SAS ENTERPRISE GUIDE 6.1 SCHEDULING New Interface

16 DEMO SAS ENTERPRISE GUIDE 6.1

17 WHAT S NEW IN SAS/STAT?

18 SAS/STAT High Performance Procedures in 12.3 HP procedures included with existing SAS 9.4 analytic products; single machine mode processing available on SAS servers and desktops. Distributed mode requires High Performance Statistics (add-on). Statistics Data Mining and Text Mining Econometrics Time Series Data Preparation HPLOGISTIC HPREG HPLMIXED HPNLIN *HPSPLIT *HPGENSLECT HPREDUCE HPNEURAL HPFOREST HP4SCORE HPDECIDE HPTMINE HPTMSCORE HPCOUNTREG HPSEVERITY HPQLIM Operations Research HPLSO Forecasting *HPFORECAST HPDS2 HPDMDB HPSAMPLE HPSUMMARY HPIMPUTE HPBIN HPCORR *New for SAS 9.4 SAS/STAT 12.3 Documentation SAS/Stat High-Performance Procedure Documentation

19 HP PROCEDURES 12.3

20 HP PROCEDURES 12.3

21 EXAMPLE HPBIN PROCEDURE proc hpbin data=ex12 numbin=5; input x1/numbin=4; input x2; ods output Mapping=Mapping; run; proc hpbin data=ex12 WOE BINS_META=Mapping; target y/level=nominal order=desc; run; HPBIN Procedure Documentation

22 EXAMPLE HPIMPUTE PROCEDURE proc hpimpute data=sampsio.hmeq out=out1; input mortdue value clage debtinc; impute mortdue / value = 70000; impute value / method = mean; impute clage / method = random; impute debtinc / method = pmedian; run; HPIMPUTE Procedure Documentation

23 HP PROCEDURES 12.3

24 HP PROCEDURES 12.3

25 EXAMPLE HPGENSELECT PROCEDURE proc hpgenselect data=getstarted; class C1-C5; model Total = C1-C5 / Distribution=Tweedie Link=Log; File='ScoringParameters.txt'; run; code HPGENSELECT Procedure Documentation

26 EXAMPLE HPSPLIT PROCEDURE proc hpsplit data=sashelp.hmeq maxdepth=7 maxbranch=2; target BAD; input DELINQ DEROG JOB NINQ REASON / level=nom; input CLAGE CLNO DEBTINC LOAN MORTDUE VALUE YOJ / level=int; criterion entropy; prune misc / N <= 6; partition fraction(validate=0.2); rules file='hpsplhme2-rules.txt'; score out=scored2; run; HPSPLIT Procedure Documentation

27 SAS/STAT 13.1 KEY FEATURES OF SAS/STAT 13.1 Bayesian choice modeling Item response theory models Competing-risk model available in PROC PHREG Interval-censored survival analysis Sensitivity analysis for missing data

28 SAS/STAT 13.1 NEW PRODUCTION PROCEDURES Three procedures become production with the 13.1 release. These include: ADAPTIVEREG procedure for fitting multivariate adaptive regression spline models QUANTLIFE procedure for quantile regression for right-censored data QUANTSELECT procedure for model selection for quantile regression models

29 SAS/STAT 13.1 UPDATES FOR TABLE ANALYSES Mid p-values are now computed for exact tests offered in the FREQ procedure (and PROC NPAR1WAY). In addition: Score confidence intervals are available for odds ratios and relative risks. Mantel-Haenszel and summary score estimates of the common proportion difference are available.

30 SAS/STAT 13.1 HIGHLIGHTS OF OTHER ENHANCEMENTS The Tweedie distribution is now supported by the GENMOD procedure (and HPGENSELECT). The elastic net method is provided in the GLMSELECT procedure. PROC MCMC has been multithreaded, providing performance boosts for models with complex likelihoods and/or many random effects. Power for PROC GLM-type MANOVA and repeated measurements is now provided in PROC GLMPOWER. PROC SURVEYMEANS produces domain quantile estimates. SAS/STAT 13.1 Product Documentation

31 WHAT S NEW IN SAS/ETS?

32 SAS/ETS What is ETS? Collection of Econometric and Time Series procedures Some tools are explanatory (Econometric: Why do things happen?) and some are predictive (Forecasting: What will happen tomorrow?) Types of tools in SAS/ETS Data Aggregation Time Series Analysis Time Series Econometrics Cross-section Econometrics Panel Data Econometrics

33 SAS/ETS WHAT S NEW IN 13.1 Binary\Censored\Truncated Regression with Endogenous Regressors Bayesian Multivariate Models Compound Distribution Models for Aggregate Loss Modeling Vector autoregression computational enhancements New features for State-Space Modeling Access to Federal Reserve Economic Data (FRED)

34 ADDITIONAL FEATURES AUTOREG: multiple structural change tests Bai and Perron (1998) COUNTREG supports STORE, RESTORE and SCORE New procedures: HPCOPULA, HPCDM, HPPANEL PANEL: system GMM estimator by Blundell and Bond (1998) SASEXFSD: Uses FactSet FASTFetch web service SEVERITY: Support for scoring functions UCM: bootstrap-based procedure for computing the standard errors of the series and component forecasts

35 SAS/ETS ECONOMETRIC RESOURCES AT SAS Forecasting and Econometrics Community at communities.sas.com Full list of product features on SAS website at support.sas.com/ets Sample programs located at

36 WHAT S NEW IN SAS ENTERPRISE MINER 13.1

37 INNOVATIVE RELEASE PROVIDES NEW DATA MINING FEATURES Machine Learning HP Random Forest HP Support Vector Machines* HP Neural HP Cluster* Scalability HP Principal Comp* HP Generalized Linear Models* HP Bayesian Networks* Time Series Data Mining (3)* Productivity Open Source Integration* Register Models* Save Data* Clustered Mid-Tier Integrated Windows Authentication SAS Enterprise Miner 13.1 * New node or procedure ** Shipped in December 2013, requires Base SAS 9.4M1

38 SCALABLE IN- MEMORY ANALYTICS SAS HIGH-PERFORMANCE DATA MINING Use all of your data: Hadoop, Pivotal (Greenplum), Oracle, Teradata Model Extensively Iteratively, Frequently Make Informed Decisions

39 COMPLETE LIST OF SAS ENTERPRISE MINER 13.1 NODES SAMPLE Append Data Partition File Import Filter Merge Sample Input Data EXPLORE Association Cluster Graph Explore Variable Clustering DMDB MultiPlot Market Basket StatExplore Link Analysis Path Analysis Variable Selection SOM/Kohonen MODIFY Drop Impute Interactive Binning Principal Components Replacement Rules Builder Transform Variables Decision Tree AutoNeural Regression Neural Network Partial Least Squares Dmine Regression DM Neural Ensemble Rule Induction Gradient Boosting LARS MBR Two Stage Model Import MODEL Incremental Response Survival Analysis Credit Scoring* TS Correlation TS Data Prep TS Dimension Reduction TS Decomp. TS Similarity TS Exponential Smoothing HP Explore HP Impute HP Regression HP Transform HP Variable Selection HP Neural HP Forest HP Decision Tree HP Data Partition HP GLM HP SVM HP Cluster HP Principal Components ASSESS Cutoff Decisions Model Comparison Score Segment Profile UTILITY Control Point End Groups Start Groups Open Source Integration Reporter Score Code Export Metadata SAS Code Ext Demo Save Data Register Metadata *Requires Credit Scoring for SAS Enterprise Miner Add-on License

40 HP NODES ENTERPRISE MINER HP Cluster HP Data Partition HP Explore HP Forest HP GLM HP Impute HP Neural HP Principal Components HP Regression HP SVM HP Text Miner HP Transform HP Tree HP Variable Selection

41 SAS ENTERPRISE MINER 13.1 HP GENERALIZED LINEAR MODEL NODE Uses the high-performance HPGENSELECT procedure to fit a generalized linear model in a threaded or distributed computing environment Several response probability distributions and link functions are available Provides model selection methods

42 SAS ENTERPRISE MINER 13.1 HP PRINCIPAL COMPONENTS AND CLUSTER NODES Perform principal component analysis for data dimension reduction, a frequent intermediate step in the data mining process Perform k-means clustering analysis in threaded and distributed computing environments, using numeric interval variables as inputs

43 ENHANCED NODES HP NEURAL Provides a User-Defined Architecture, which gives users more control over the construction of the neural network. Users can specify the number of hidden layers, the number of hidden neurons, and associated activation functions for each layer. Users can configure Input and Target Standardizations, Target Error, and Activation Functions.

44 SAS ENTERPRISE HP SUPPORT VECTOR MACHINE NODE MINER 13.1 Enables the creation of linear and nonlinear support vector machine models Constructs separating hyperplanes that maximize the margin between two classes Enables use of a variety of kernels: linear, polynomial, radial basis function, and sigmoid function. The node also provides interior point and active set optimization methods.

45 ENHANCED NODES HP FOREST Allows users to optionally perform variable selection based on either OOB Average Error for interval targets or OOB marginal reduction for class targets.

46 SAS ENTERPRISE MINER 13.1 HIGH-PERFORMANCE BAYESIAN NETWORK PROCEDURE Analyzes different types of Bayesian network structures, including naive, tree-augmented naive (TAN), Bayesian networkaugmented naive (BAN), parentchild Bayesian network, and Markov blanket Performs efficient variable selection through independence tests, and automatically selects the best model from the specified parameters

47 NEW NODE OPEN SOURCE INTEGRATION NODE (R SUPPORT) Enables users to integrate R code (supervised and unsupervised models) inside a SAS Enterprise Miner process flow diagram Provides flexibility to include R code within a data mining flow, using SAS Enterprise Miner for data prep, R for modeling, and then SAS Enterprise Miner for deployment Includes R models in assessment of models generated by SAS Enterprise Miner and, in some R-generated PMML (Predictive Model Markup Language) cases, corresponding SAS DATA step scoring code

48 SAS ENTERPRISE MINER 13.1 TIME SERIES DATA MINING NODES Time Series Dimension Reduction node extracts features from each time series and reduces the dimension of time Time Series Correlation node helps you perform correlation and cross-correlation analyses Time Series Decomposition node enables you to perform seasonal decomposition of time series

49 NEW NODES REGISTER MODEL NODE Enables users to register segmentation, classification, or prediction models to the SAS Metadata Server. Models registered in metadata can then be accessed by Model Manager, Enterprise Miner, Web Services, etc Information about inputs, outputs, targets and SAS score code is registered to metadata.

50 NEW NODES SAVE DATA NODE Provides users a simple way to save training, validation, test, score, or transaction data from a SAS Enterprise Miner path to a user-defined path, or a previously defined SAS library. Data can be exported as a SAS table, a JMP file, an Excel 2010 spreadsheet, a flat CSV file, or a tab-delimited file.

51 ENHANCED NODES DECISION TREE NODE Enables users to import a previously created model and apply this model to new data. Adds High-Performance support for models that have interval targets. Adds the criterion for determining the associated splitting rules can be based on a reduction in variance, or by F test. Adds more flexibility and control exists when handling missing values. This means that the user can map missing values to either the largest branch, the most correlated branch, or to a separate branch.

52 SAS ENTERPRISE MINER RESOURCES

53 DEMO SAS ENTERPRISE MINER

54 SAS ENTERPRISE MINER RESOURCES What s New in SAS Enterprise Miner 13.1 (SAS Talk) SAS Predictive Analytics and Data Mining SAS Enterprise Miner Documentation SAS Enterprise Miner Online Community (Please Join!)

55 SAS ENTERPRISE MINER YOUTUBE VIDEOS What's new in SAS Enterprise Miner 13.1 YouTube HP GLM Node SAS High-Performance Text Mining Incremental Response Modeling Link Analysis in SAS Enterprise Miner 12.3 Survival Node in SAS Enterprise Miner

56 SNEAK PEAK AT WHAT S COMING SOON!

57 WHAT S COMING WHERE THERE S A a 13.2 will follow! Stay tuned for the 2014 summer release of SAS/STAT 13.2, including: Additional features for missing data analysis Additional updates to survival data analysis software More ordinal data analysis capabilities in PROC LOGISTIC And more, much more..

58 WHAT S COMING SAS STUDIO

59 IT S HERE SAS STUDIO

60 WHAT S COMING SAS IN-MEMORY ANALYTICS FOR HADOOP

61 WHAT S COMING SAS VISUAL STATISTICS

62 WHAT S COMING SAS ENTERPRISE MINER

63 WHAT S COMING ANALYTIC DIRECTIONS / ROADMAP More frequent analytic release schedule High Performance Computing Continued development in Bayesian Analysis & Modeling Analysis of Time-To-Event Data Missing Data Methods Modeling of Complex Data Current Directions in SAS/STAT Software Development Paper

64 QUESTIONS? Thank you for your time and attention! sas.com