Azure Machine Learning

This is a very simple example based on the Tour tutorial given in the studio but as you can see, it takes in a data set, splits the data, trains the model, scores the model, then you can evaluate the model which is the data and that will be your output. In the Evaluate Model box, there is an option where you can visualize your data after the algorithm has finished running and you can see the results of the algorithm.

Use case for a business

As of 06/27/2017, there are 905 case studies available through the use of Azure Machine Learning from industries such as banking, education, healthcare, manufacturing, media, retail, and a few more.

So coming back to our earlier example of items, let’s be more specific. Let’s say I, as a retailer sell mobile phones. In my data set, I placed sales from the past 2 years also including when they were sold etc. As part of my data set, I can also add data such as specs like megapixels of the camera, screen resolution, etc. Since I’m a retailer selling phones, I release a new version of the phone every year. So we have 2 data sets:

⦁ A phone that was sold in year one

⦁ A new phone that was sold in year two

I can then place both of these data sets in the algorithm. And automatically, it will tell me the sales / comparison between the two. But, let’s say I want to know if my sales will increase or stay the same if I added some new features or updated some features on my phone. I can use one of the available modules in the studio to make that difference happen and I can get a prediction of what those sales might look like based on my past 2 data sets. Similarly, I can even predict sales for the next 10 months and use this information to design my next new phone or upgrade the specs of my existing models with the help of machine learning.

Is Azure Machine Learning my only option?

With the increasing interest in cloud based technologies, Azure Machine Learning is not your only option. While they all offer the same functionality, they work a little bit different from an user interface (UI) standpoint and how algorithms are designed.

Amazon also makes it easy to build out complex ML algorithms easily through models

IBM Watson  offers key features such as simple model creation and deployment of models

BigML was one of the first contenders to bring machine learning for end users and make it easy

Google offers an open source software library called TensorFlow that is used to code Machine Intelligence

But which one should you pick?

 You can’t really go wrong with either choice, they all offer free learning to use their product. What it really comes down to is preference and what kind of community you want to be involved in. Because they are different companies, they all have different communities and maybe you are a bigger fan of Microsoft than Amazon. So as a user who can’t decide, your best bet would be to explore the products, see how they work then make your decision.

With Microsoft as a new entrant in this technology, users can rest assured that Azure machine learning will have a lot to offer over the next few years as the product improves and works better with businesses worldwide.