Recommended
More Related Content
Similar to Serving predictive models with Redis
Similar to Serving predictive models with Redis (20)
Recently uploaded
Recently uploaded (20)
Serving predictive models with Redis
- 1. Home of Redis Serving Predictive Models with Redis Tague Griffith Head of Developer Advocacy
- 2. 2 Topics • Introductions • Why Machine Learning • What is Apache Spark • Redis-ML
- 4. 4 Who I am • Head of Developer Advocacy for Redis Labs • Developer and architect turned Evangelist • Infrastructure and Distributed Systems • Large Scale Redis Systems • Former: Apple, Netscape, Yahoo/Flickr, GoPro • Focus on the Open Source Community • Education and Support • Nurture and grow the entire community
- 5. 5 Redis Labs – Home of Redis Founded in 2011 HQ in Mountain View CA, R&D center in Tel-Aviv IL The commercial company behind Open Source Redis Provider of the Redis Enterprise (Redise) technology, platform and products
- 6. 6 Redise Cloud Private Redis Labs Products Redise Cloud Redise Pack ManagedRedise Pack SERVICES SOFTWARE Fully managed Redise service in VPCs within AWS, MS Azure, GCP & IBM Softlayer Fully managed Redise service on hosted servers within AWS, MS Azure, GCP, IBM Softlayer, Heroku, CF & OpenShift Downloadable Redise software for any enterprise datacenter or cloud environment Fully managed Redise Pack in private data centers && &
- 7. 7 REmote DIctionary Server Strings Hashes Lists Sets Bitmaps Hyperlog- logs Sorted Sets Geospatial Bitfield
- 8. 8 A Quick Recap of Redis Key "I'm a Plain Text String!" { A: “foo”, B: “bar”, C: “baz” } Strings / Bitmaps / BitFields Hash Tables (objects!) Linked Lists Sets Sorted Sets Geo Sets HyperLogLog { A , B , C , D , E } [ A → B → C → D → E ] { A: 0.1, B: 0.3, C: 100, D: 1337 } { A: (51.5, 0.12), B: (32.1, 34.7) } 00110101 11001110 10101010
- 9. 9 Redis Main Differentiations Simplicity (through Data Structures) Extensibility (through Redis Modules) Performance ListsSorted Sets Hashes Hyperlog-logs Geospatial Indexes Bitmaps SetsStrings Bit field
- 11. 11 Teaching a computer by example to learn an algorithm that is too complex to program
- 12. 12 Machine Learning Problems Pick One of a Set • Spam Detection • Manufacturing defect detection • Handwriting analysis • Decision Trees/Forests • Naïve Bayes Score or Rank • Recommendations • Likelihood of Purchase • Linear Regression • Logistic Regression Classification Regression Group Similar • Find Similar Items • Customer segmentation • Cohort detection • K-Means • Hierarchical Clustering Clustering
- 13. 13 Supervised Learning – Training Spam Classifier Mail Spam Mail MailMail Mail Spam SpamSpam Mail MailSpam Spam Mail Spam
- 14. 14 Deploying a Spam Classifier 14 Spam Mail Spam Spam
- 15. 15 How do we Build these Boxes ¯_(ツ)_/¯
- 16. 16 Typical Spark Application Structure 16 Spark Training Data is loaded into Spark Model is saved in files File System Custom Server Model is loaded to your custom app Serving Client Client App
- 17. 17 Building high performance and reliable services are hard, isn't there something we can deploy
- 18. Redis - ML
- 19. 19 Redis Modules • Any C/C++ program can now run on Redis • Use existing or add new data-structures • Enjoy simplicity, infinite scalability and high availability while keeping the native speed of Redis • Can be created by anyone New Capabilities New Commands New Data Types
- 20. 20 Redis-ML: Predictive Model Serving Engine • Predictive models as native Redis types • Perform evaluation directly in Redis • Store training output as “hot model” Spark Training Data loaded into Spark Model is saved in Redis-ML Redis-ML Serving Client Client App Client App Client App Any Training Platform
- 21. 21 Redis ML Module Redis Module Tree Ensembles Linear Regression Logistic Regression Matrix + Vector Operations More to come...
- 22. 22 Random Forest Model • A collection of decision trees • Supports classification & regression • Splitter Node can be: ◦ Categorical (e.g. day == “Sunday”) ◦ Numerical (e.g. age < 43) • Decision is taken by the majority of decision trees
- 23. 23 Classic Tree Problem: Titanic Survival YES Sex = Male ? Age < 9.5? Sibps > 2.5? Survived Died SurvivedDied NO • Passenger Data encoded as feature vectors • ML Algorithm learns the tree rules • ID3, CART (RPART), etc. • Tree rules used to infer results
- 24. 24 Titanic Survival: Random Forest YES Sex = Male ? Age < 9.5? *Sibps > 2.5? Survived Died SurvivedDied NO YES Country= US? State = CA? Height> 1.60m? Survived Died SurvivedDied NO YES Weight< 80kg? I.Q<100? Eye color =blue? Survived Died SurvivedDied NO Tree #1 Tree #2 Tree #3
- 25. 25 Who Would Survive the Titanic John: • Male, 34, • Married w/ 2 kids (Sibps=3) • New York, USA • 1.78m, 78kg • 110 iq • Blue eyes Mathew: • Male, 6 • 3 Sisters (Sibps=3) • New York, USA • 1.06m, 22.7 kg • 100 iq • Brown eyes Let's use our forest to find out
- 26. 26 Redis: Forest Data Type Add nodes to a tree in a forest: Perform classification/regression of a feature vector: ML.FOREST.ADD <forestId> <treeId> <path> [ [NUMERIC|CATEGORIC] <splitterAttr> <splitterVal> ] | [LEAF] <predVal> ML.FOREST.RUN <forestId> <features> [CLASSIFICATION|REGRESSION]
- 27. 27 Real World Challenge • Ad serving company • Need to serve 20,000 ads/sec @ 50msec data-center latency • Runs 1k campaigns → 1K random forest • Each forest has 15K trees • On average each tree has 7 levels (depth)
- 28. 28 Ad Serving costs: Homegrown v. Redis Homegrown 1,247 x c4.8xlarge 35 x c4.8xlarge Cut computing infrastructure by 97% 28
- 29. 29 Redis ML with Spark ML Random Forest; 1,000 forests @ 15,000 trees Classification Time Over Spark 13x Faster
- 33. 33 Step 1: Get The Data Download and extract the MovieLens 100K Dataset The data is organized in separate files: • Ratings: user id | item id | rating (1-5) | timestamp • Item (movie) info: movie id | genre info fields (1/0) • User info: user id | age | gender | occupation Our classifier should return the expected rating (from 1 to 5) a user would give the movie in question
- 34. 34 Step 2: Transform 34 The training data for each movie should contain 1 line per user: • class (rating from 1 to 5 the user gave to this movie) • user info (age, gender, occupation) • user ratings of other movies (movie_id:rating ...) • user genre rating averages (genre:avg_score ...) Run gen_data.py to transform the files to the desired format
- 35. 35 Step3: Train and Load to Redis // Create a new forest instance val rf = new RandomForestClassifier().setFeatureSubsetStrategy("auto").setLabelCol("indexedLabel").setFeat uresCol("indexedFeatures").setNumTrees(500) ….. // Train model val model = pipeline.fit(trainingData) ….. val rfModel = model.stages(2).asInstanceOf[RandomForestClassificationModel] // Load the model to redis val f = new Forest(rfModel.trees) f.loadToRedis(”movie-10", "127.0.0.1")
- 36. 36 Step 4: Execute inference in Redis Redis-ML + Spark Training Client App
- 37. 37 Summary • Train with Spark, Serve with Redis • 97% resource cost serving • Simplify ML lifecycle • Redise (Cloud or Pack): ‒Scaling, HA, Performance ‒PAYG – cost optimized ‒Ease of use ‒Supported by the teams who created Spark and Redis Spark Training Data loaded into Spark Model is saved in Redis-ML Redis-ML Serving Client Client App Client App Client App +
- 38. 38 Where to Find Me @tague https://github.com/tague tague@redislabs.com