Machine learning allows us to build predictive analytics solutions of tomorrow - these solutions allow us to better diagnose and treat patients, correctly recommend interesting books or movies, and even make the self-driving car a reality. Microsoft Azure Machine Learning (Azure ML) is a fully-managed Platform-as-a-Service (PaaS) for building these predictive analytics solutions. It is very easy to build solutions with it, helping to overcome the challenges most businesses have in deploying and using machine learning. In this presentation, we will take a look at how to create ML models with Azure ML Studio and deploy those models to production in minutes.
2. About Me
Microsoft, Big Data Evangelist
In IT for 30 years, worked on many BI and DW projects
Worked as desktop/web/database developer, DBA, BI and DW architect and developer, MDM
architect, PDW/APS developer
Been perm employee, contractor, consultant, business owner
Presenter at PASS Business Analytics Conference, PASS Summit, Enterprise Data World conference
Certifications: MCSE: Data Platform, Business Intelligence; MS: Architecting Microsoft Azure
Solutions, Design and Implement Big Data Analytics Solutions, Design and Implement Cloud Data
Platform Solutions
Blog at JamesSerra.com
Former SQL Server MVP
Author of book “Reporting with Microsoft SQL Server 2012”
4. Fully
managed
Integrated Best in Class
Algorithms
Deploy in
minutes
No software to install,
no hardware to manage,
and one portal to view
and update.
Simple drag, drop and
connect interface for
Data Science. No need
for programming for
common tasks.
Built-in collection of
best of breed
algorithms. Support for
R and Python for
extensibility.
Operationalize models
with a single click.
Monetize in Machine
Learning Marketplace.
10. This is Karl.
Karl owns a company that
operates vending machines in
Washington state.
His job is to make sure that his 100
vending machines are selling drinks
& obtaining revenue.
Karl wants revenue to always
be high & his business to
be profitable
11. Sadly, vending machine will
occasionally break & may take up
to 7 days to fix, thus hurting sales.
To eliminate this occurrence, Karl must maintain
operations & figure out the best way to utilize
resources in order to optimize revenue.
12. Azure Cloud Services + Machine Learning to the Rescue!
1. Which Machines Have Failed?
2. Which Machines Will Soon Fail?
13. • Damage is reported by customer
or during weekly restocking routes
• Technician must be scheduled
to investigate
• Process take up to 8 days to fix
a broken machine
• Sensor data is used to monitor
cooler condition in real-time
• Broken coolers are identified
at time of failure
• Lost sales remain due to
maintenance lead teams
(parts & repair technicians)
• Azure ML predicts where, when,
& what failures will occur based
on sensor data
• Spare parts & repairs can be
scheduled before machines shut
down leading to no lost sales
CURRENT SCENARIO REAL-TIME SENSORS SENSORS&MACHINELEARNING
Days: Days:Days:
21. • Accessible through a web browser, no software
to install;
• Collaborative work with anyone, anywhere via
Azure workspace;
• Visual composition with end2end support for
data science workflow;
• Best in class ML algorithms; Immutable library of
models, search discover and reuse;
• Extensible, support for R & Python;
• Rapidly try a range of features, ML algorithms
and modeling strategies
22. Cortana Intelligence Suite
Integrated as part of an end-to-end suite
Action
People
Automated
Systems
Apps
Web
Mobile
Bots
Intelligence
Dashboards &
Visualizations
Cortana
Bot
Framework
Cognitive
Services
Power BI
Information
Management
Event Hubs
Data Catalog
Data Factory
Machine Learning
and Analytics
HDInsight
(Hadoop and
Spark)
Stream Analytics
Intelligence
Data Lake
Analytics
Machine
Learning
Big Data Stores
SQL Data
Warehouse
Data Lake Store
Data
Sources
Apps
Sensors
and
devices
Data
23. Event Hub
Stores Streaming
Data
Stream Analytics
processes events as they
arrive in the EventHub
AML Model
Web Service
BES endpoint
Power BI / D3
Dashboard
Data for
Real-time
Processing
Aggregations
External Data Azure Services
Azure SQL
Contains Historical Energy
Consumption Data
Real time
data stats
Azure Data Factory
Pipeline invokes AML
Web Service
RealTimeBatch
Example Architecture
Real Time
Telemetry
Data
Azure Data Factory
Pipeline Moves Data
Batch updates
of predictions
AML Model
Web Service
RRS endpoint
24. Real Time Energy
Consumption Data
(Public Source)
Event Hub
Stores Streaming
Data
Stream Analytics
processes events as they
arrive in the EventHub
AML Model
Web Service
BES endpoint
Power BI / D3
Dashboard
Data for
Real-time
Processing
Data Stream
Job
Hourly Prediction
Updates
External Data Azure Services
Copy to Azure SQL
for batch predictions
Scrape Data
5 mins
Azure WebJob
Runs jobs to scrape
data from public source
Azure SQL
Contains Historical Energy
Consumption Data
Real time
data stats
Azure Data Factory
Pipeline invokes AML
Web Service
RealTimeBatch
25. Q & A ?
James Serra, Big Data Evangelist
Email me at: JamesSerra3@gmail.com
Follow me at: @JamesSerra
Link to me at: www.linkedin.com/in/JamesSerra
Visit my blog at: JamesSerra.com (where this slide deck is posted via the
“Presentations” link on the top menu)
Notes de l'éditeur
Machine learning allows us to build predictive analytics solutions of tomorrow - these solutions allow us to better diagnose and treat patients, correctly recommend interesting books or movies, and even make the self-driving car a reality. Microsoft Azure Machine Learning (Azure ML) is a fully-managed Platform-as-a-Service (PaaS) for building these predictive analytics solutions. It is very easy to build solutions with it, helping to overcome the challenges most businesses have in deploying and using machine learning. In this presentation, we will take a look at how to create ML models with Azure ML Studio and deploy those models to production in minutes.
Fluff, but point is I bring real work experience to the session
My goal is to give you a high level overview of all the technologies so you know where to start Make you a hero
Advanced Analytics, or Business Analytics, refers to future-oriented analysis that can be used to help drive changes and improvements in business practices. It is made up of four phases:
Descriptive Analytics: What is generally referred to as “business intelligence”, this phase is where a lot of digital information is captured. Then this big data is condensed into smaller, more useful nuggets of information, creating an understanding of the correlations between those nuggets to find out why something is happening (“Diagnostic Analytics”). In short, you are providing insight into what has happened to uncover trends and patterns. An example is Netflix using historic sales and customer data to improve their recommendation engine.
Predictive analytics: Utilizes a variety of statistical, modeling, data mining, and machine learning techniques to study recent and historical data, thereby allowing analysts to make predictions, or forecasts, about the future. In short, it helps model and forecast what might happen. For example, taking sales data, social media data, and weather data to forecast the product demand for a certain region and to adjust production. Or you can use predictive analytics to determine outcomes such as whether a customer will “leave or stay” or “buy or not buy.”
Prescriptive analytics: Goes beyond predicting future outcomes by also suggesting actions to benefit from the predictions and showing the decision maker the implications of each decision option. Prescriptive analytics not only anticipates what will happen and when it will happen, but also why it will happen. The output is a decision using simulation and optimization. In short, it seeks to determine the best solution or preferred course of action among various choices. For example, airlines sift thought millions of flight itineraries to set an optimal price at any given time based on supply and demand. Also, prescriptive analytics in healthcare can be used to guide clinician actions by making treatment recommendations based on models that use relevant historical intervention and outcome data.
Our embrace of open source continues with our GA release by now adding to our full support of R with new first class support for SQLite and Python. Just as you can with R, you can drop your custom Python code directly into ML Studio.
You can then combine your custom work with our world-class algorithms such as learning with counts, which allows you to digest terabytes of data – truly a Big Data solution in the cloud.
And, as I’ve mentioned, no other vendor offers the ability to deploy with one click – you can literally see the button here. It is simply powerful to be able to put your models into production and increase their usability for businesses, the community and the world.
And we’re enabling the community of Azure ML users by offering an in-product Gallery where you can discover solutions built by others, drop those directly into your workspace for experimentation and learning, as well as easily share your own work in the gallery with the same ease of use you experience with web service production. You can try this out by logging on to azure.com/ml and clicking “get started” to try our perpetually free version.
Microsoft’s Big Data vision in the cloud is to enable organizations to solve large, complex problems end-to-end, from storing and managing TBs of data without investing in hardware and software, to seamless integration with the 1 billion users of Excel. As part of this vision, Microsoft offers Azure Machine Learning, designed to democratize the complex task of advanced analytics.
Advanced analytics is using products like Azure Machine Learning to find new and actionable insights that traditional approaches to business intelligence are unlikely to discover. An easy way to think about this is thinking about a dashboard. Today when confined by only BI tools without a connection to machine learning, it is solely the job of the human looking at the spreadsheet to gain insights and react to the data. But a human can only consume so many variables. A computer, on the other hand, can consume a great deal more variables to provide much deeper insight on the data. Humans can then react to the data to make decisions that drive competitive advantage, as well as program the computer further to recognize important patterns in the future. This is why we say beyond business intelligence – machine intelligence.
The accessibility of our solution starts with set up. Previously you needed to provision your workspace on-premises for machine learning, also thinking about server space and a host of other considerations. Today you can get started with just a browser. With only an Azure subscription, you can take advantage of the full functionality of Azure Machine Learning within minutes. Taking a test drive is even easier, click Get Started off azure.com/ml and with simply a Microsoft ID you’re off to the races.
Another limit with other machine learning solutions are siloed environments that only allow for one programming language or make changing from one algorithm to another time consuming and complex. With Azure ML, you can experience the power of choice. That choice expands to language, with both Python and R being first class citizens of Azure ML, or algorithm. You can choose from hundreds of algorithms, including business-tested ones running our Microsoft businesses today. And swapping out algorithms to land on the right one for you is done with a click. Additionally you can drop in custom R and Python code – your “special sauce” – and mix and match that with the other options in the tool.
Most revolutionary of all you can deploy solutions in minutes as a web service, which is simply a url which can connect to any data, anywhere – including on-premises or in another cloud environment. The ability to put a model into production almost immediately, as well as revise it easily, is unique to Microsoft and allows companies to stay on top of the changing business landscape more effectively than is offered by any other provider today.
We even take that a step further, allowing model developers to connect to the world with our Machine Learning Marketplace, where they can publish finished solutions and APIs with their own brand and business model. Developers can also discover machine learning solutions there without any machine learning skills needed – the data science is inside. Check it out at https://datamarket.azure.com/.
Offer structures: A la carte, Data Intensive, Analytics Intensive, Stream Intensive, All-inclusive