AZURE.COM/MLCosting SharePoint Projects with Azure Machine LearningJohn Timney
JOHN TIMNEY Microsoft SharePoint MVP & 2010 & 2013, 2016 TAP member 25 years+ of ugly IT Managing Enterprise Architect Primarily worked in large organisations, on large projects
IT Services Agency, Syntegra, BT PLC Capgemini Hewlett Packard Enterprise
Specialise in large scale Enterprise Strategy, Architecture, Assurance and Governance – usually with SharePoint/Office 365 and Azure at scale
Co- authored a few books on various SharePoint, JAVA and .NET subjects North East Administrator for the SharePoint UK User Group
Recently completed Assurance for a 300,000 seat, 1,000,000 device Azure and SharePoint / 0365D Hybrid Cloud implementation.
What is Machine Learning?
Azure MachineHow it works
Learning:
Azure Machinein action
Learning
Get Inspired
Contents
What isMachine Learning?Predictive computingsystems become smarterwith experience
WHAT IS IT?
Machine learning is a type of artificial intelligence (AI) that provides computers with the ability to learn without being explicitly programmed.
Machine learning focuses on the development of computer programs that can teach themselves to grow and change when exposed to new data.
Stanford Uni built an algorithms that learned to recognize cats from sampling millions of (cute) cat images on the internet without having any prior concept of a “cat.”
Why Learn, why Predict?Learn it when you can’t code it (e.g. speech recognition)
Learn it when you can’t scale it (e.g. recommendations, Spam detection)
Learn it when you have to adapt/personalize (e.g.
predictive typing)
Learn it when you can’t track it (e.g. robot control)
The United States Postal ServiceCan process over 150 billion pieces of mail per year —far too much for efficient human sorting.
of
But as recently as 1997, only 10%hand-addressed mail wassuccessfully sorted automatically.
of
The challenge in automation isenabling computers to interpretendless variation in handwriting.F
A LESSON IN HISTORYThe evolution of Machine Learning
Microsoft
&Machine
Learning1
5years
of
realizing
innovation
2014
Microsoft
launches
Azure MachineLearnin
g
2012
Successful,
real-time, speech-
to- speech
translation
2010
Microsoft Kinect can
watch users
gestures
2008
Bing Maps ships with
ML traffic-
prediction service
2005
SQL Serverenable
sdata
mining
2004
Microsoft search
engine built with machine learning
1999
Computers work on users behalf,
filtering junk email
Machine Learning is pervasive through Microsoft Products
ExpensiveSiloeddata
Break awayfrom industry limitations
FragmentedtoolsDeploymentcomplexity
Huge set-up costs of tools, expertise, and compute/storage capacity create unnecessary barriers to entry
Siloed and cumbersome data managementrestricts access to data
Complex and fragmented tools limit participation in exploring data and building models
Many models never achieve business value due to difficulties with deploying to production
Machine Learning TodayHard to Reach Solutions
http://www.gartner.com/newsroom/id/3114217
SMART MACHINE TECHNOLOGIES ARE COMING
By 2018, 40 percent of outsourced services will leverage smart machine technologies, rendering the offshore model obsolete for competitive advantage
Smart Machine Technologies Will Render the Offshore Model Obsolete for Competitive Advantage
‘virtual talent.’ is coming - It's faster, cheaper and more predictable
Big shifts in the vendor landscape - product vendors offering "business process as a service" or cognitive business offerings.
Start to build the capability to analyze, rethink, reimagine and recalibrate your sourcing portfolio
http://www.gartner.com/newsroom/id/3207317
THE FEEDING OF THE 5000
“We are on the cusp of a paradigm shift in computing that is unlike anything we have seen in decades,”
Microsoft CEO Satya Nadella
HOW AML WORKS
Enabling custom predictive analytics solutions at the speed of the market
CORTANA INTELLIGENCE SUITE
Packaging a raft of Azure services into a single suite
Business Intelligence, Big Data, and Advanced Analytics service offerings, machine learning, digital assistance, IM and Big data – - all interconnected
The Environments
The Team
DevelopersML API service
Data Scientists
ML Studio
Azure Ops TeamAzure Portal
How it works
Azure Portal
Azure Ops Team
Desktop Data
Azure Storage
HDInsightML Studio
Data Scientist
Azure Portal & ML API service
Azure Ops Team
ML API service Jupyter/R/Python/F# Developer Web Apps
Mobile Apps PowerBI/Dashboards
One Solution for Machine learning – Raw data to Prediction
Azure Portal
Azure Ops Team
ML Studio
Data Scientist
Azure Portal & ML API service
Azure Ops Team
ML API service Developer
Business users easily access results:from anywhere, on any device
Desktop Data
Azure Storage
HDInsightML Studioand the Data Scientist
• Access and prepare data• Create, test and train models• Collaborate• One click to stage for
production via the API service
Azure Portal & ML API service
and the Azure Ops Team• Create ML Studio workspace• Assign storage account(s)• Monitor ML consumption• See alerts when model is ready• Deploy models to web service
ML API service and the Developer• Tested models available as an url that can be called from any end point
Web Apps Mobile Apps PowerBI/Dashboards
Fully managedNo software to install, no hardware to manage, and one portal to view and update
Easy to use
Simple drag, drop and connect interface you can access and share from anywhere
Tested solutionsAccess to sample experiments, tested algorithms, support for custom R, and over 350R packages
Deploy in minutesTooled for quick deployment, hand-off and updates
Solutions as the Market demands them
Directly In the Azure portal
Azure MachineLearning in actionA simple no-code / no calculation SharePoint cost Model
https://azure.microsoft.com/en-gb/documentation/articles/machine-learning-algorithm-cheat-sheet/
Use linear regression when you want a very simple model for a basic predictive task.
Linear regression tends to work well on high-dimensional, sparse data sets lacking complexity.
https://studio.azureml.net/
https://manage.windowsazure.com/
THE ALGORITHMS Linear Regression is a machine learning algorithm used to
predict a numeric outcome Train Model algorithm uses historical data from which to learn
patterns it can base predictions on. Score Model generates just the predicted numeric value we
are seeking. Evaluate Model measures the accuracy of the regression
model. A new Apply Transformation model is created at RUN, and
some webservices added http://www.johntimney.com/wp-content/uploads/2015/11/spreadsheet.zip
COEFFICIENT OF DETERMINATION
Mean absolute error (MAE) measures how close the predictions are to the actual outcomes; thus, a lower score is better.
Root mean squared error (RMSE) creates a single value that summarizes the error in the model. By squaring the difference, the metric disregards the difference between over-prediction and under-prediction.
Relative absolute error (RAE) is the relative absolute difference between expected and actual values; relative because the mean difference is divided by the arithmetic mean.
Relative squared error (RSE) similarly normalizes the total squared error of the predicted values by dividing by the total squared error of the actual values.
Mean Zero One Error (MZOE) indicates whether the prediction was correct or not. In other words: ZeroOneLoss(x,y) = 1 when x!=y; otherwise 0
Coefficient of determination, often referred to as R2, represents the predictive power of the model as a value between 0 and 1.
https://studio.azureml.net
Azure MachineLearningThe Potential is Endless
AMELIAShe is designed to communicate like a human using “natural language”.The company says the workforce of the future will have to include both human and virtual employees in order to compete in the digital economy.It is said to be 60 percent cheaper than using a human worker, Any questions Amelia cannot answer will be referred to a human colleague.
detection &
Spamfiltering
Frauddetection
Anomalydetection
Equipment monitoring
Recommendations
Forecasting
Churn analysis
Ad targeting
Image
classification
Using Past Data to predict the Future
Imagine what machine learning could do for your business?
WHEN IT ALL GOES WRONG Facebook is well-known for its
amazing yet scary data mining algorithms that suggest you friends. It compares all the information you offer to suggest you connections. For example, you see people from your high school, from the same living area, your interests, mutual friends, and even people from other networks as well in suggestions.
ANY QUESTIONS….?