Azure ML - a Mindshift for Business Intelligence professionals
This article was commissioned by Jen Stirrup a SQL Server fan and Microsoft MVP with a passion for Business Intelligence and Data Virtualisation. On the 12th of November, Jen will be speaking on the topic of ‘Azure ML – a Mindshift for Business Intelligence professionals’ at Future Decoded which is a free event which will see keynotes from Brian Cox, Sir Nigel Shadbolt, Or Arbel and Michael Taylor, followed by eight individual speaker tracks that discuss different topics. Be sure to register before the day is sold out, and we hope to see you there!
There is a paradigm mindshift in the world of data. Businesses are starting to understand the value in their data. Suddenly, data has become accessible and almost ‘cool’. Readers may be familiar with sabermetrics, which is the empirical analysis of baseball using statistics. Film fans will have seen this concept in the film Moneyball, starring Brad Pitt. Even Fox are joining in the data fun, offering a separate broadcast of a Saturday night’s NLCS Game 1 matchup between the Giants and Cardinals with a focus on statistics, statistics and graphics. Although baseball statistics are fun, how can Machine Learning be used for businesses, and how is it relevant to Business Intelligence?
Firstly, what is machine learning? It is the application of maths for the purpose of process of pattern discovery in data. Machine Learning has been more commonly used historically as a domain which is more specific to maths, and its application to the development of algorithms which are implemented for the purposes of getting patterns out of data. Data Mining is more a business term, where the objective is to make a business decision dependent on the outcome. Perhaps because Data Mining is a business-focused term, it is more commonly used than Machine Learning.
As there is a growth in data, there is a growth of interest in Machine Learning. Why is machine learning becoming a focal point of interest again? Organisations have an issue of increased data, and businesses are starting to take an interest in what can be done with the data. Machine learning has also become interesting to a wider range of people. It is attracting an organic mix of skill sets, including data scientists, business analysts, and even business leaders. Many people use ML algorithms every day, and may not even realise it. For example, if you use Bing search and the auto-fill appears, this is the result of an ML algorithm which is predicting your results. Spam filters in email also use ML algorithms. If you’ve been internet shopping recently, and had ‘recommendations’ based on your browsing history, then this is the result of a ML algorithm crunching through your preferences to produce results. Other, less visible examples include credit card fraud detection, market segmentation, and scientific analysis such as bioinformatics.
David McCandless is quoted recently as saying that ‘data is the new soil’ – something to be mined, and made into something of value. Businesses are starting to look at having a data culture in their organisations, whereby the data is perceived as the lifeblood of the organisation. What does this mean for the Business Intelligence professional? It can mean using existing skill sets, since Machine Learning involves a multi-faceted sets of skills and the renewed interest means that Business Intelligence professionals will start to think about how to take the next step away from Business Intelligence and insights into data that happened until this point, and moving forward into doing advanced analytics to make predictions and patterns with the data that they have so carefully curated in the organisations. Machine Learning, in the first wave, was the domain of mathematicians and computer science experts. Now, it is becoming more accessible to everyone, and Business Intelligence professionals are in a great place to adopt ML technologies.
There are also a number of external factors which are levellers in the business world, and make machine learning accessible to businesses who may not have been able to access it previously. Machine learning has now become part of the estate that is accessible and available to everyone. Cost is being removed as a barrier to Machine Learning due to cheaper commodity computing, and improved accessibility to cloud computing. Solutions such as Microsoft’s Azure Machine Learning offer an lower risk, operational cost which is attractive to businesses, who may not otherwise have the capital expenditure to lay out for computers and software licenses, which were previously the domain of larger organisations with deeper pockets.
Organisations who are ‘newbies’ to the world of data insights don’t need to fear the complexity of trying to generate their own algorithms, or take on expensive consultants to do the work for them. Simply put, it is a safe path for organisations to simply try out Machine Learning for themselves, in order to see how it works for them. Businesses do not like risk or uncertainty, but they do need to find a way to make the most of their data. As with most things, you need to make a change to get a chance, and trying out Azure Machine Learning which are low cost, low risk. Businesses can see if there are insights in their data, without a huge outlay in terms of capital expenditure or up-front investment in data science skills to create advanced algorithms from scratch. You, the Business Intelligence professional, can be part of that data story.
One key word in data is self-service; it is a thread that runs through many solutions. Microsoft’s Power BI is advertised as a self-service Business Intelligence solution. However, how can you support the self-service business user to go from self-service Business Intelligence, to self-service analytics which uses Machine Learning? In other words, how do we take the Excel Pro business user, and move them towards cloud-based Machine Learning? Fortunately, there are a plethora of training materials in Machine Learning, which are in response to a desire for people to learn more about it. For example, PASS are setting up a Machine Learning Virtual Chapter, serving free online tutorials aimed at Azure Machine Learning. Coursera support Stanford University in offering a free online Machine Learning series, and free in-person events such as SQLSaturday London BA Edition offer sessions in Machine Learning. In Azure Machine Learning itself, there are a range of hands-on tutorials and videos for people to try it out. Again, it is a safe way to learn more about this fascinating technology.
Azure ML also allows for R coding in the cloud. What is R? R is the world’s most widely used data analysis software, which is used by over 2 million data scientist, statisticians and analysts to analyse data. It is one of the world’s most powerful statistical programming language. It is open-source, flexible, extensible and comprehensive for productivity. R is on the leading edge of analytics research, and it used in many universities worldwide. Incorporating R into Azure ML means that it is accessible to people who want to use R to analyse data, and the power of the cloud means that they can use it on larger data sets. R is no ‘flash in the pan’ and new graduates are often familiar with R, meaning that there is a pool of people who already know how to use it. R has a thriving open-source community, and it has its own subset of Google – Rseek.org– which means that it is easy to find resources on it. Outside of Azure ML, R can be used to create beautiful and unique data visualisations - as seen in New York Times, Twitter and Flowing Data website.
R is the point at which business analysts and data scientists can speak to one another in a common canvas. Putting it into Azure ML was a smart move; it is a familiar software, and it has a heritage of authenticity in the statistics sphere generally. It can also be used in conjunction with Microsoft Power BI, because the output of R can be visualized in Microsoft Excel and Power BI.
To summarise, Azure ML becomes a part of the data ecosystem in an organization, which requires a mindshift from working with Business Intelligence to more advanced analytics. The next step in this series is to provide some examples whereby we can show how we can adopt a mindshift from Business Intelligence to advanced analytics using Azure ML.