Azure Stream Analytics is a service that enables you to write SQL like queries on-top of a stream and it is currently in preview here https://azure.microsoft.com/en-us/documentation/services/stream-analytics/ however how do you put an architecture together that contains the building blocks necessary to enable this.

One of my colleagues put together a new reference architecture that describes how this can be done you can find it here:

https://aka.ms/realtimeanalytics

The table of contents looks like this:

Chapter 1: What is Machine Learning?
Chapter 2: Planning for Machine Learning
Chapter 3: Working with Decision Trees
Chapter 4: Bayesian Networks
Chapter 5: Artificial Neural Networks
Chapter 6: Association Rules Learning
Chapter 7: Support Vector Machines
Chapter 8: Clustering
Chapter 9: Machine Learning in Real Time with Spring XD
Chapter 10: Machine Learning as a Batch Process
Chapter 11: Apache Spark
Chapter 12: Machine Learning with R

I read the book on my Kindle and I did not do the labs but because of illustrations and code listings I had no problems following along.

What I thought of the book

I found the book an interesting read I liked how the author laid out the subjects. He gave a lot of examples and also went through some of the math behind the algorithms. However I got the feeling that he had not made up his mind on what the book should be. From chapter 9 and onward the book was much more a Hadoop/Spark/Spring XD book than anything else and the only examples used was variations on word count with some twitter analytics thrown in. In his introduction the author talks about that machine learning are systems that learn and improve with experience yet the text analysis that he shows is simple word matches instead of real NLP implementations.

Who I think should read it

Developers that want to get an introduction to machine learning and the tools used. Most of the examples are in Java so it is good if the reader can follow along with code in that language.

The author is an individual that have a lot of background in the area himself having started 3 companies in the data analysis space. The table of contents looks like this:

Chapter 1: Chris Wiggins, The New York Times
Chapter 2: Caitlin Smallwood, Netflix
Chapter 3: Yann LeCun, Facebook
Chapter 4: Erin Shellman, Nordstrom
Chapter 5: Daniel Tunkelang, LinkedIn
Chapter 6: John Foreman, MailChimp
Chapter 7: Roger Ehrenberg, IA Ventures
Chapter 8: Claudia Perlich, Dstillery
Chapter 9: Jonathan Lenaghan, PlaceIQ
Chapter 11: André Karpištšenko, Planet OS
Chapter 12: Amy Heineike, Quid
Chapter 13: Victor Hu, Next Big Sound
Chapter 14: Kira Radinsky, SalesPredict
Chapter 15: Eric Jonas, Neuroscience Research
Chapter 16: Jake Porway, DataKind

The interviews follow the same pattern with open ended questions on who the person are, the company they represent what they do and what tools they use. Rarely there is any follow-up questions asked.

What I thought of the book

I found the book a bit long, and sometimes tedious to read. In my opinion the most interesting part to be what reading and resources they recommended. I also found some of the questions really strange like what was the first data set they worked on, why is that an interesting question to me at least that has not been life changing. Sometimes I gott a bit frustrated that there are no follow-up questions, rarely the question why is used, it is always about the what and how. The people interviewed are well selected and interesting but after reading the book I cannot wonder how the book would have turned out if the author would have been an experienced journalist sine doing good interviews is extremely hard.

Who I think should read it

Someone with interest in the start-up community in US. You also should not be annoyed with the fact that the book sometimes feels like reading a who-is-who in data science in Silicon Valley and New York, both the interviewees and the interviewer are guilty of name dropping.

The first one is Predictive Analytics with Microsoft Azure Machine Learning: Build and Deploy Actionable Solutions in Minutes by Roger Barga, Valentine Fontama and Wee Hyong Tok released on Apress.

I have read the Kindle version of the book and found it to really work well in that format. The book consist of 8 chapters that tries to teach how to apply Azure Machine Learning to solve various business problems.

The table of contents look like this:
Chapter 1: Introduction to Data Science
Chapter 2: Introducing Microsoft Azure Machine Learning
Chapter 3: Integration with R
Chapter 4: Introduction to Statistical and Machine Learning Algorithms
Chapter 5: Building Customer Propensity Models
Chapter 6: Building Churn Models
Chapter 7: Customer Segmentation Models
Chapter 8: Building Predictive Maintenance Models

The first section (chapter 1-3) serves as an introduction to the area and the service and its features. Section 2 (chapter 4) contains information on the algorithms applied in the rest of the book. The last section, section 3 (chapter 5-8) shows how you can apply algorithms to solve real problems.

What I thought of the book

I do not claim to be a data scientist or an expert in machine learning but I have been using Azure Machine Learning in real life. Still I found the book valuable, I learnt some new techniques and got some valuable background information. What I really found valuable was the notes and references to other books and papers especially when they did not come from Wikipedia. What could have been covered to greater extent would have been data preparation. How do you prepare a dataset for let's say churn analysis? In the book the examples often started with a sample set that had been pre-prepared earlier.

Who I think should read it

Someone with a background in data science or BI who wants to learn Azure ML but wants something more than the tutorials available in the help pages for Azure ML.

Second Packt publishing is offering books including mine called "Getting Started with SQL Server 2012 Cube Development" for just $5 as a holiday promotion. Find out more about the offer and order a bunch of books at https://bit.ly/1C4FAaQ

Part of the role is also to be involved in spreading the gained knowledge to the larger community. I have been involved in the creation of a white paper on big data patterns. This work is something that has been driven out of our Patterns & Practices team and hopefully it contains useful information to everyone that is considering a big data solution. You can find the paper here https://aka.ms/pnp-bigdata-guidance

By the way Packt Publishing is celebrating 10 years as a publisher and is offering $10 off on my book (and all other books) you can find more about the offer at https://bit.ly/1iMqxxB, the offer is valid until the 5th of July so if you want some summer reading be sure to pick up some books.

Now you have the chance of buying this book and getting one for free due to the fact that my publisher have published their 2000th title around IT books. The offer is available until 26^thof March 2014.

You can find the offer here https://bit.ly/1j26nPN

The book that I have written covers the entire BI development process and this is the table of contents:

Chapter 1: Self-service Business Intelligence, Creating Value from Data

• Identifying common user requirements for a BI system
• Creating a flexible reporting solution
• Getting user value through self-service reporting
• Summary

Chapter 2: Installing SSAS and Preparing for Cube Development

• Understanding the BI semantic model architecture
• Choosing the deployment mode
• Actions
• Aggregations
• Custom assemblies
• Custom rollups
• Distinct count
• Linked objects
• Many-to-many relationships
• Parent-child hierarchies
• Translations
• Writeback
• Tool support
• Installing Analysis Services 2012 in multidimensional mode
• Hands-on steps for installing Analysis Services
• Hands-on steps for attaching the sample database
• Starting SQL Server Data Tools for the first time
• Choosing the correct project
• Hands-on steps for creating your first cube project
• Navigating the project environment
• Solution Explorer
• Properties window
• Build menu
• Hands-on steps for configuring your cube project
• Summary

Chapter 3: Creating Your First Multidimensional Cube

• An introduction to data warehousing
• Understanding data sources
• Creating the data connection
• Understanding the data source view
• Creating a new data source view
• Adding objects to the data source view
• Extending the data source view
• Understanding dimensions
• Creating the Date dimension
• Preparing the Date dimension for end users
• Renaming attributes
• Creating a hierarchy and attribute relationships
• Processing the dimension to review the results
• Creating the first cube
• Creating the Fact Internet Sales cube
• Creating additional dimensions and adding them to the cube
• Summary

Chapter 4: Deploying and Processing Cubes

• Deploying objects to Analysis Services
• Deploying the FirstCube project to the server
• Deploying projects using advanced deployment strategies
• Processing objects in Analysis Services
• Processing the FirstCube project
• Using advanced processing options when processing cubes
• Scheduling processing
• Building a SSIS package to control processing
• Troubleshooting processing errors
• Partitioning cubes to speed up processing
• Adding partitions to the FirstCube cube
• Configuring storage modes in Analysis Services
• MOLAP
• ROLAP
• HOLAP
• Proactive caching
• Adding a ROLAP dimension to the FirstCube cube
• Summary

Chapter 5: Querying Your Cube

• Understanding multidimensionality
• Writing MDX queries using Management Studio
• Connecting to the cube using Management Studio
• Understanding the anatomy of an MDX query
• The FROM clause
• The WHERE clause
• The query axis clause
• Important query concepts
• Unique names
• Finding unique names of objects
• Tuples
• Sets
• Calculated members
• Named sets
• Functions
• Using Excel as the query tool against cubes
• Connecting Excel to the FirstCube cube
• Advanced Excel features
• Named sets
• Calculated measures
• Calculated members
• Using Reporting Services together with your cube
• Summary

Chapter 6: Adding Functionality to Your Cube

• Adding measures
• Adding additional measure groups
• Adding dimensions to the FirstCube project
• Adding referenced dimensions
• Adding many-to-many dimensions
• Adding dimensions with parent-child hierarchies
• Adding calculations to cubes
• Simple calculated measures
• Adding advanced calculated measures
• Adding calculated members
• Time and date calculations
• Key Performance Indicators
• Adding perspectives to simplify cube browsing
• Adding translations to support global implementations
• Extending the cube with custom actions
• Building budget solutions using writeback
• Summary

Chapter 7: Securing Your Cube Project

• Understanding role-based security
• Adding users to the fixed server role
• Adding custom roles to the database
• Securing objects in Analysis Services
• Adding dimension security to the FirstCube project
• Securing measures
• Implementing data security
• Testing data security
• Enable Visual Totals to restrict what a user can see
• Understanding security when a user belongs to several roles
• Implementing dynamic data security
• Summary

Chapter 8: Using Aggregations to Performance Optimize a Cube

• Understanding cube aggregations
• Adding aggregations to improve performance
• Running the aggregation wizard
• Adding aggregations manually
• Using usage-based optimization to optimize the cube
• Summary

Chapter 9: In-memory, the Future

• Understanding tabular models
• Installing a tabular mode instance
• Creating a tabular mode project
• Defining the workspace server
• Connecting to the data source
• Adding tables to the data model
• Adding data from other data sources
• Working with partitions in tabular models
• Creating calculations
• Adding calculated columns to the tabular model
• Creating calculated measures in the tabular model
• Creating advanced calculations
• Specifying other types of tables and columns
• KPIs
• Adding hierarchies
• Sorting data based on other columns
• Hiding columns and tables from the end users
• Creating perspectives
• Adding security to your in-memory model
• Optimizing the tabular model for performance
• Querying tabular models
• Summary

Chapter 10: Cubes in the Larger Context

• Using Microsoft frontend tools to query Analysis Services
• Developer-focused tools
• Using Reporting Services to query your cube
• SharePoint PerformancePoint Services
• Self-service tools
• Using Excel as a self-service tool against Analysis Services cubes
• Using Excel PowerPivot to allow user-created models
• SharePoint Excel Services
• Introducing Power View – an analytical tool
• Third-party tools
• Summary

To buy it and get the offer available until 26^th of March 2014 visit the following site https://www.packtpub.com/getting-started-with-sql-server-2012-cube-development/book

First of all go to https://manage.windowsazure.com and login with your credentials. In the portal then click the NEW button:

You will get into a small wizard where you can select the type of machine that you want to create,

Select one of the SQL 2012 Enterprise versions on either Windows 2008 R2 or Windows 2012.

Specify the size of the machine, machine name as well as the username and password that you wish to use for the account. In the next dialog you need to fill in more information about the machine.

In the last step you need to create the valid endpoints needed to connect to SSAS and SQL Server

More information about this can be found in the following articles:

https://technet.microsoft.com/en-us/library/ms174937.aspx

https://msdn.microsoft.com/en-us/library/windowsazure/jj992719.aspx#bkmk_endpoints

Now you can just provision your machine and connect to it through the remote desktop client. To do so first click on the newly created Virtual machine and then  click the CONNECT button in the management portal

The next step is to allow for connections to Analysis Services and SQL Server through the firewall. When you have logged on to the computer the Initial Configuration Tasks wizard will start, ensure that you first of all have downloaded and applied all the updates available for your machine. After that you can click the Configure Windows Firewall button.

In the Windows Firewall window click the Advanced Settings option.

Right-click on the Inbound Rules option and select New Rule

Configure the rule by following these steps. First specify that it should be using a Program rule.

In the next dialog specify the program path to be %ProgramFiles%\Microsoft SQL Server\MSSQL11.MSSQLSERVER\MSSQL\Binn\sqlservr.exe, you may need to change the instance name if you have a named intstance.

In the next dialog choose that it should Allow the connection.

Specify that the rule should be valid for or network profiles.

Name the rule MSSQL and click Finish.

Perform the same steps for a rule that allows you to connect to Analysis Services. This time specify the path to the program to be %ProgramFiles%\Microsoft SQL Server\MSAS11.MSSQLSERVER\OLAP\bin\msmdsrv.exe, and the name of the rule to be MSSAS.

You will now have two inbound rules that allows connections to SSAS and SQL.

Next you can download the sample databases from the following link:

https://msftdbprodsamples.codeplex.com/releases/view/55330

What you need is the following downloads:

Attach the AdventureWorksDW2012 database to your server, if you don’t know how you can follow the tips in this blog https://www.mssqltips.com/sqlservertip/2638/attach-sample-database--adventureworks-in-sql-server-2012/

Since I want to test how anonymous access works to my system over the internet I want to allow everyone the rights to read the sample cube. This is in general not recommended but here is how you can do it.

In the sample cube create a new role by right-click on Roles and select New Role.

Call the role Anonymous Access or something similar, make sure that the role have have Read definition enabled.

Add the NT AUTHORITY\ANONYMOUS LOGON as a member of the role.

Grant Read access to the cubes.

Now process and deploy your cube. What you then need to do is to allow anonymous access on the instance level as well. Connect to Analysis Services using Management Studio and right-click on the server and select Properties. Click on General and change the Security\RequireClientAuthentication to false. Then click OK.

Now in your connection string you need to specify Impersonation level=Anonymous to be able to connect to the server.  The simplest way to create such a connection is to create a new text file and rename it to <something>.udl. This will allow you to use the Data Link wizard to configure the connection string.

I usually right-click on my desktop and select New and Text Document

Then rename it to .udl 

Click OK in the following warning dialog

Double click on the file and then specify the Microsoft OLE DB Provider for Analysis Services 12.0 as the provider.

On the Advanced tab specify that you want to use Anonymous as the impersonation level.

Now you can click the connection tab and specify the server name and cube name

Click OK to save the UDL file, this file can now be used in Excel when defining the connection to the cube.

You can find the book titled Getting Started with SQL Server 2012 Cube Development here https://amzn.com/B00F60Z69E

From the 1st of November I will also change my job at Microsoft. I will be joining the EMEA part of the Data Insights Center of Excellence team as a Solutions Architect. This team is incubated by the Server and Tool Business and Enterprise Services in support of a companywide strategic initiative, driving Data Insights solutions (Big Data/Business Analytics/Business Intelligence) at customers around the world. In the new role I will hopefully be able to write more blog posts about Big Data and Business Intelligence.