- An error has occurred, which probably means the feed is down. Try again later.
High level features of Azure Machine Learning
Hi All
In our previous article we “Starting with Azure Machine Learning” we created a free Microsoft Machine Learning account. If you haven’t done that, please create your own free ML account so you can practise and understand this article further. If you have an account already created and we are all good to go.
Azure Machine Learning Studio is a GUI-based integrated development.
Our next step is to understand very high level Azure ML options. Below are high level feature in Azure ML followed by understanding of each high level feature:
Projects: As you can see in the screenshot, Projects is a newly added feature to Microsoft Azure Machine Learning. An individual can work on multiple dataset, Notebook and experiment for different clients/projects. Unless we use a proper naming convention, it was difficult to understand which dataset is associated to which project/client. To resolve this issue, Projects was introduced so that we can consolidate all information related to single project in same location. I would say Projects was a very small change to Azure ML but it has a very big Impact.
Experiments: This is the brain and heart of Azure ML. Experiments is where we create new models and train our model on smaller data. We can iterate on our model design, editing the experiment and running it until it gives us the results we looking for. Experiments in Azure ML is a showcase for simple experiments that make it easier for someone who wants to start exploring machine learning, through datasets. We would be spending nearly 70 % of our time in Experiments by using different features like data format conversions, data input and output, feature selection, etc. Below is screenshot of same:
Web Services : When our model is ready, we can publish it as a web service so that others can send it new data and get predictions in return. This is a very important feature in Azure ML and we will discuss a lot about this feature in our upcoming articles.
NoteBooks : As you can see in the screenshot, Notebooks is a newly added feature to Microsoft Azure Machine Learning. As we have already seen and discussed, Azure ML is GUI based, drag and drop simple to use ML platform however we have lot of other platform which are non GUI base. Azure ML Notebooks is create to encourage those who still want to continue writing code. We can create a machine learning model in a notebook by writing the code and implementing the machine learning API.
Datasets: The first and the foremost thing we need in Machine Learning is data. Azure ML offers several sample datasets which are part of Machine Learning Studio. We ca use it or we have option to import data from many sources like from our local machine. Very common dataset formats are GenericCSV, Attribute-Relation File Format (ARFF), GenericTSV, DataTableDotnet, etc
Trained Models : We can say that trained Model is a part of web services. When our model is ready, we can publish it as a web service so that others can send it new data and get predictions in return. Similarly, I can use this trained models for my other dataset without publishing it and use it within same workspace. This is a very important feature in Azure ML and we will discuss a lot about this feature in our upcoming articles.
Settings: Workspace name, storage, account information, users, data gateways and Authorization tokens are options available in settings. This is just like any other settings available in applications.
I will explain uses of each and every High level feature and talk more in details about them in my upcoming articles.
Happy Learning!!