Fault Detection using Advanced Analytics at CERN's Large Hadron Collider: Too Hot or Too Cold BIWA Summit 2016

Transcription

1 Fault Detection using Advanced Analytics at CERN's Large Hadron Collider: Too Hot or Too Cold BIWA Summit 2016 Mark Hornick, Director, Advanced Analytics January 27, 2016

2 Safe Harbor Statement The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, and timing of any features or functionality described for Oracle s products remains at the sole discretion of Oracle. 2

3 Oracle s Advanced Analytics Multiple interfaces across platforms SQL, R, GUI, Dashboards, Apps Users R programmers Data / Business Analysts Business Analysts/Mgrs Domain End Users R Client SQL Developer/ OBIEE Applications Oracle Data Miner Platform Hadoop ORAAH Parallel, distributed algorithms Oracle Database Enterprise Edition Oracle Advanced Analytics - Database Option SQL Data Mining & Analytic Functions + R Integration for Scalable, Distributed, Parallel in-database ML Execution Oracle Cloud Oracle Database 12c 3

4 Scaling R to Big Data Immediate access to database and Hadoop data from R Eliminate need to request extracts from IT/DBA Process data where they reside minimize or eliminate data movement through data.frame proxies Scalability and Performance Use parallel, distributed algorithms that scale to big data on Oracle Database Leverage powerful engineered systems and RAC environments to build and score models on billions of rows of data or build millions of models in parallel from R Ease of deployment Using Oracle Database, place R scripts immediately (no need to recode) in production via SQL Use production quality infrastructure without custom plumbing or extra complexity Process support Maintain and ensure data security, backup, and recovery using existing processes Store, access, manage, and track analytics objects (models, scripts, workflows, data) in Oracle Database 4

5 Scaling R to Big Data Oracle R Enterprise, part of Oracle Advanced Analytics option Use Oracle Database as HPC environment Use in-database parallel and distributed machine learning algorithms Manage R scripts and R objects in Oracle Database Integrate R results into applications and dashboards via SQL Client R Engine ORE packages Oracle Database User tables In-db stats SQL Interfaces SQL*Plus, SQLDeveloper, Database Server Machine 5

6 Oracle R Enterprise three key areas Transparency Layer Standard R syntax to manipulate data in-database via proxy objects (ore.frame) Overloads R functions and translates functionality to SQL Rich set of transformations, statistical functions, visualizations Predictive Analytics In-database ODM algorithms exposed via R interface with auto data preparation ORE-specific parallel, distributed algorithms Embedded R Execution Invoke user-defined R functions from R and SQL at database server machine Use CRAN packages in user-defined R functions Execute in data-parallel and task-parallel manner to leverage powerful machines like Exadata Return structured data, images, XML via SQL invocation for easy integration with applications R Script Repository and Datastore as R object repository 6

7 Scalability through proxies with function overloading In-database aggregation no data movement Oracle Distribution of R version (--) -- "Full of Ingredients" > aggdata <- aggregate(ontime_s$dest, + by = list(ontime_s$dest), + FUN = length) > class(aggdata) [1] "ore.frame" attr(,"package") [1] "OREbase" > head(aggdata,4) Group.1 x 1 ABE ABI 34 3 ABQ ABY 10 Oracle Advanced Analytics ORE Client Packages Transparency Layer Oracle SQL select DEST, count(*) from ONTIME_S group by DEST ONTIME_S In-db Stats 7

8 Scalable Machine Learning Algorithms ORE parallel distributed model (e.g., Linear Regression) using embedded R engines Oracle Distribution of R version (--) -- "Full of Ingredients" > options(ore.parallel=4) > lm_mod <- ore.lm(arrdelay ~ DISTANCE + DEPDELAY, data=ontime_s) Oracle Advanced Analytics ORE Client Packages Predictive Analytics > summary(lm_mod) Call: ore.lm(formula = ARRDELAY ~ DISTANCE + DEPDELAY, data = ONTIME_S) Residuals: Min 1Q Median 3Q Max Coefficients: Estimate Std. Error t value Pr(> t ) (Intercept) 2.254e e e-05 *** DISTANCE e e < 2e-16 *** DEPDELAY 9.625e e < 2e-16 *** --- Residual Signif. codes: standard 0 error: *** on ** degrees * of 0.05 freedom (4785 observations deleted due to missingness) Multiple R-squared: , Adjusted R-squared: F-statistic: 3.497e+05 on 2 and DF, p-value: < 2.2e-16 ONTIME_S 1 2 Parallel ORE Framework 3 extproc extproc extproc external procedure Oracle R Distribution Parallel Oracle ore.lm R Distribution Compute R Oracle Parallel Packages R Distribution ore.lm Compute Parallel R Oracle Packages ore.lm R Distribution Compute Parallel R Packages Compute Engine 4 8

9 Oracle R Enterprise Predictive Analytics algorithms in-database plus open source R packages for algorithms in combination with embedded R data- and task-parallel execution Classification Decision Tree Logistic Regression Naïve Bayes RandomForest Support Vector Machine Regression Linear Model Generalized Linear Model Multi-Layer Neural Networks Stepwise Linear Regression Support Vector Machine New in ORE 1.5 Clustering Hierarchical k-means Orthogonal Partitioning Clustering Attribute Importance Minimum Description Length Anomaly Detection 1 Class Support Vector Machine Market Basket Analysis Apriori Association Rules Feature Extraction Nonnegative Matrix Factorization Principal Component Analysis Singular Value Decomposition Time Series Single Exponential Smoothing Double Exponential Smoothing 9

10 Internet of Things Use Case: Energy Demand Leveraging ORE Embedded R Execution Model each customer s usage to understand behavior and predict individual usage and overall aggregate demand 200 thousand households, each with a utility smart meter 1 reading / meter / hr 200K x 8760 hrs / yr 1.752B readings 3 years worth of data 5.256B readings Each customer has readings If each model takes 10 seconds to build, hrs (23.2 days) with 128 DOP 4.3 hrs

11 Scalable Analysis Model Building Smart meter scenario Oracle Database + ORE Data c1 c2 ci cn R Datastore R Script Repository f(dat,args, ) f(dat,args, ) f(dat,args, ) f(dat,args, ) f(dat,args, ) { R Script build model Model c1 Model c2 Model ci Model cn }

12 Build models, partitioned on CUST_ID, and store in database ore.groupapply (CUST_USAGE_DATA, CUST_USAGE_DATA$CUST_ID, function(dat, ds.name) { cust_id <- dat$cust_id[1] mod <- lm(consumption ~. -CUST_ID, dat) mod$effects <- mod$residuals <- mod$fitted.values <- NULL name <- paste("mod", cust_id,sep="") assign(name, mod) ds.name1 <- paste(ds.name,".",cust_id,sep="") ore.save(list=paste("mod",cust_id,sep=""), name=ds.name1, overwrite=true) TRUE }, ds.name="mydatastore", ore.connect=true, parallel=128 ) 12

13 Next Generation Workflow for data collection / model building Oracle Database with OAA / ORE Big Data SQL R Client OAA/ORE R integration SQL Client 13

14 Next Generation Workflow for data collection / model building Oracle Database with OAA / ORE Big Data SQL R Client OAA/ORE R integration SQL Client 14

15 Next Generation Streaming Machine Learning Architecture Input event stream Model transfer via Oracle Stream Explorer Data prep, model import, scoring function creation Data Scoring Prediction stream Oracle Database Data Training Data Model Building ore.groupapply + {pmmll} package Oracle R Enterprise Datastore ore.save R Script Repository 15

16 Check out this ORE session for more details on streaming machine learning Thursday at 9:50 AM, Room 104 Thursday at 11:55 AM, Room

17 Oracle Advanced Analytics On Premise or Cloud 100% Compatibility Enables Easy Coexistence and Migration CoExistence and Migration Same Architecture On-Premise Same Analytics Same Standards Oracle Cloud Transparently move workloads and analytical methodologies between On-premise and public cloud 17

18 To Learn More about Oracle s R Technologies 18

19