We all know life can get hectic. Here at Pragmatic Works, we're no different. But one of our goals is to learn something new about Azure every day, as things are constantly changing and being updated. Many people are still learning all the amazing things they can do within the Azure cloud and we want to help. Our posts in our Azure Every Day series are a great way to learn more about Azure each week.
If you’re like many of our customers, you’re looking to learn more about advanced analytics, machine learning and other data science type topics. In today’s post, I’d like to discuss what we do to help our customers get to where they want to be in these areas.
When we start out with a customer, one goal is to help them have a growth strategy and to understand how to move from what they’re doing today to what they’re looking at tomorrow. I like to call this ‘future proofing’. Yes, I know nothing is future proof, but it’s important to plan so you don’t design yourself into a corner.
When it comes to data science, machine learning and such, what we typically see are scenarios where a customer may have already built or are currently building a data warehouse solution in Azure. If they followed our guidance, their data lands in Azure Blob Storage or the Data Lake and is staged there to allow for easy access by other tooling.
If a client has no experience with machine learning or data science and they’re trying to get their feet wet and understand if it will be beneficial to them, we want to give them the opportunity to do that at a low cost. Azure Machine Learning Studio provides that kind of opportunity.
For newcomers, we suggest diving into Azure Machine Learning by working in the Studio, which has a simple drag and drop interface, to build a solution that is straightforward and easy to handle. This is a perfect starting point for someone at an entry level to begin to understand how it fits into their world.
In doing it this way, it allows us to isolate algorithms that could work, find gaps in the models but still test and verify that we’re on the right track, without having to learn a bunch of R or Python packages to be able to move to the next step. And it’s beneficial to take the time to learn and understand the capabilities in Azure Machine Learning and the algorithms you use to build your models.
Then when you’re ready for the next step, if your solution works well, you can productionize that and use it in a way that you can attach to web services and other tools; maybe you don’t go beyond that if it fits your needs. As you start to advance, you’ll probably be looking at R or Python. Once you get into that realm you’ll need to begin looking into other tools and options.
One thing to consider at that point is using Azure Databricks. With Databricks you can create notebooks that allow you to build out the next steps in your solution. This is important so you don’t bind yourself into a corner. If your data is in Azure storage, you’ve refined it and got some work done there, you can move forward with more advanced techniques using Azure Databricks and the machine learning capabilities there.
As you continue to move forward, you may find other open source tooling and things like using R Studio with SQL Server. There are many options available to allow you to expand on what you’re learning and what you’ve done.
My point is it’s always a good thing to start simple and move to the complex. Then you can begin to make decisions such as: do I need to ramp up my staff on Python or R? What are we already working in and of those things, what can I leverage? But at the same time not losing the momentum around data science and machine learning and what you’re trying to understand and learn.
One of the cool things about Azure Machine Learning Studio (and machine learning in general) is there are lots of pre-built solutions that you can just plug your data in and try it. So, if you’re new in your data science journey and want to try something low risk and low cost but gain an understanding of the potential benefit of using machine learning and similar tooling, Azure Machine Learning is a great place to start.
You can start there, knowing that you can jump into Databricks or other advanced techniques without losing the effort you’ve already put in; you may need to do some rework but that’s how it goes when you move to the next tier of things.
If you’d like to learn more about Azure Machine Learning Studio, or machine learning in general, as well as the how data science can be beneficial to your business, we can help. Our team of experts can help you every step of the way in your machine learning or Azure journey. Click the link below or contact us to start a conversation today!