Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Predicting NBA winners with Postgres and Watson Studio

135 views

Published on

Learn how to clean data, store and retrieve from PostgreSQL, and build a machine learning model and deploy it to the web

Published in: Data & Analytics
  • Be the first to comment

  • Be the first to like this

Predicting NBA winners with Postgres and Watson Studio

  1. 1. @rajrsingh Data science and machine learning with Postgres on the IBM Cloud Predicting NBA winners with Postgres and Watson Studio Raj Singh Developer Advocate IBM Watson & Cloud IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation photo from Wikimedia Commons
  2. 2. @rajrsingh Featured ServicesThis is the IBM Cloud Compute Network Storage Management Security Data Analytics AI IoT Mobile Developer Tools Blockchain Integration Migration Private Cloud VMware IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation 2 Watson ML, Deep Learning, SpeechóText, Visual Recognition Object Storage, Block Storage, CDN Cloud Foundry, Containers, Cloud Functions, Bare Metal Watson Studio, Spark, Streaming Analytics Relational, NoSQL, Migration
  3. 3. @rajrsingh ML choices in Watson Studio IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation 3 • R • Scikit-learn • Tensorflow • WatsonML/SparkML ü deploy to cloud!
  4. 4. @rajrsingh web pipeline The ML Process IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation training data + Vector Assembler array of columns Estimation Model regression, decision tree, random forest, etc. fitted model = fitted model = p r e d i c t i o n s • p r e d i c t i o n s • p r e d i c t i o n s • p r e d i c t i o n s • test data scoring model = p r e d i c t i o n s • p r e d i c t i o n s • p r e d i c t i o n s • p r e d i c t i o n s • real-time data _____ _____ _____
  5. 5. @rajrsingh What’s your record? Home or away? How “hot” are you? IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation 5
  6. 6. @rajrsingh web pipeline The ML Process IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation training data + Vector Assembler [homeWins, awayWins, homeLast5, awayLast5] Estimation Model logistic regression fitted model = fitted model = p r e d i c t i o n s • p r e d i c t i o n s • p r e d i c t i o n s • p r e d i c t i o n s • test data online model = p r e d i c t i o n s • p r e d i c t i o n s • p r e d i c t i o n s • p r e d i c t i o n s • real-time data 80% 20%
  7. 7. @rajrsingh 1. Code in Jupyter Notebooks 2. Choose key variables 3. Store & Retrieve data with PostgreSQL 4. Load into Spark DataFrame 5. Fit Linear Regression model with SparkML 6. Deploy to Cloud 7. Predict future games ML with PostgreSQL to Predict NBA outcomes IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation 7
  8. 8. @rajrsingh Demo NBA Explore-Clean https://gist.github.com/rajrsingh/62082835fdef01e770024140559cdd24 NBA Machine Learning https://gist.github.com/rajrsingh/8dee6303e3c94f9593499ed0b8812b7a IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation 8photo by Keith Allison https://www.flickr.com/photos/keithallison/24689492152/in/photostream/
  9. 9. @rajrsingh Learn More q Watson Studio q ibm.co/watsondataplatform q IBM Watson Data Lab q medium.com/ibm-watson-data-lab q ibm-watson-data-lab.github.io q GitHub q github.com/ibm-watson-data-lab q PixieDust q ibm.co/pixiedust qRaj Singh q rrsingh@us.ibm.com q @rajrsingh on Twitter IBM Watson & Cloud / April, 2018 / © 2018 IBM Corporation 9

×