Azure Machine 
Learning
Agenda 
About me 
Introduction to Azure Machine Learning 
Differences between Azure ML and SSAS Data Mining 
Demo 
• Azure ML 
• Consume ML Webservice from Power Query 
Quick intro to other relevant tools in Azure
About me 
David Bojsen dlb@kapacity.dk 
36 years old, married, 2 children (7 and 17) 
8 years of experience with SQL Server and Microsoft BI 
11 years of experience with C# and .NET 
BI Manager at TELMORE since April 2009 
Senior Business Analytics Architect at Kapacity since April 2013 
BI Developer – NOT data scientist
Introduction to Azure Machine Learning 
Azure Cloud Service 
Prebuilt standard models for data discovery and prediction (unsupervised 
learning or supervised learning) 
Framework for building new and/or selling models 
Allow you to build a model using R or/and standard algorithms 
About finding an algorithm/model for determining an outcome (well-defined 
problem – e.g. given a set of features what is the probability of x, what is the 
predicted price of selling a house, what classification does an ”entity” belong 
to) 
Trained AND scored based on a dataset 
Inputs is a predefined set of features that can be used to predict an outcome
Machine Learning Algorithms
Azure ML vs. SSAS Data Mining 
Azure ML 
Pay as you Go 
Fixed set of algorithms 
extensible by using R 
Easy visual interface 
Cloud (first) 
Batch / online requests 
SSAS Data Mining 
Free (With SQL Server) 
Fixed set of algoritms 
(somewhat outdated) 
Integrated in Visual Studio 
On-premise 
Batch requests 
Good for data discovery (thru 
Excel add-in)
Extensability 
R 
• Access to over 400 of the most popular CRAN packages 
• Your own custom code 
Python 
• Python Tools for Visual Studio 
(Check Azure ML Blog) 
• Not yet integrated in ML Studio
DEMO 
Building an experiment to test/train/select a model 
Building an experiment to publish a model 
Testing model manually 
Consuming a model from Power Query
Other relevant services in Azure 
Azure Stream Analytics 
• Think Stream Insight In The Cloud 
Azure Data Factory 
• Think Integration Services In The Cloud (for cloud and on-prem) 
Azure Service Bus 
Azure Event Hubs 
Azure Intelligent Systems Service 
• All related to Internet Of Things
Azure Stream Analytics
Azure Data Factory
Internet Of Things 
Data 
”Things” 
”Devices” 
ISS Agent
Want to know more? 
• Azure Machine Learning 
• Service - http://azure.com/ML 
• Documentation – http://aka.ms/MLDS 
• Pricing – http://aka.ms/MLPricing 
• Studio – http://studio.azureml.com 
• FAQ – http://aka.ms/MLFAQ 
• MarketPlace – http://aka.ms/MLMarket 
• Blog – http://aka.ms/MLBlog 
• Azure Stream Analytics – http://azure.microsoft.com/en-us/services/stream-analytics/ 
• Azure Data Factory – http://azure.microsoft.com/en-us/services/data-factory/ 
• http://www.internetofyourthings.com/

MSBIP møde nr. 25 - Azure ML

  • 1.
  • 2.
    Agenda About me Introduction to Azure Machine Learning Differences between Azure ML and SSAS Data Mining Demo • Azure ML • Consume ML Webservice from Power Query Quick intro to other relevant tools in Azure
  • 3.
    About me DavidBojsen dlb@kapacity.dk 36 years old, married, 2 children (7 and 17) 8 years of experience with SQL Server and Microsoft BI 11 years of experience with C# and .NET BI Manager at TELMORE since April 2009 Senior Business Analytics Architect at Kapacity since April 2013 BI Developer – NOT data scientist
  • 4.
    Introduction to AzureMachine Learning Azure Cloud Service Prebuilt standard models for data discovery and prediction (unsupervised learning or supervised learning) Framework for building new and/or selling models Allow you to build a model using R or/and standard algorithms About finding an algorithm/model for determining an outcome (well-defined problem – e.g. given a set of features what is the probability of x, what is the predicted price of selling a house, what classification does an ”entity” belong to) Trained AND scored based on a dataset Inputs is a predefined set of features that can be used to predict an outcome
  • 5.
  • 6.
    Azure ML vs.SSAS Data Mining Azure ML Pay as you Go Fixed set of algorithms extensible by using R Easy visual interface Cloud (first) Batch / online requests SSAS Data Mining Free (With SQL Server) Fixed set of algoritms (somewhat outdated) Integrated in Visual Studio On-premise Batch requests Good for data discovery (thru Excel add-in)
  • 7.
    Extensability R •Access to over 400 of the most popular CRAN packages • Your own custom code Python • Python Tools for Visual Studio (Check Azure ML Blog) • Not yet integrated in ML Studio
  • 8.
    DEMO Building anexperiment to test/train/select a model Building an experiment to publish a model Testing model manually Consuming a model from Power Query
  • 9.
    Other relevant servicesin Azure Azure Stream Analytics • Think Stream Insight In The Cloud Azure Data Factory • Think Integration Services In The Cloud (for cloud and on-prem) Azure Service Bus Azure Event Hubs Azure Intelligent Systems Service • All related to Internet Of Things
  • 10.
  • 11.
  • 12.
    Internet Of Things Data ”Things” ”Devices” ISS Agent
  • 13.
    Want to knowmore? • Azure Machine Learning • Service - http://azure.com/ML • Documentation – http://aka.ms/MLDS • Pricing – http://aka.ms/MLPricing • Studio – http://studio.azureml.com • FAQ – http://aka.ms/MLFAQ • MarketPlace – http://aka.ms/MLMarket • Blog – http://aka.ms/MLBlog • Azure Stream Analytics – http://azure.microsoft.com/en-us/services/stream-analytics/ • Azure Data Factory – http://azure.microsoft.com/en-us/services/data-factory/ • http://www.internetofyourthings.com/