Microsoft initially launched the service under the name of Azure Machine Learning; however, they later changed it to Azure Machine Learning Studio in line with its core capabilities of being able to manage the entire process using the Studio itself (Visual Workbench). It's a managed service that you can use to create, test, operate, and manage predictive analytic solutions in the cloud. The core value proposition of this service is that you do not necessarily need to be a data scientist in order to build machine learning models as it provides multiple sample datasets and analysis modules, which can be used together to create your machine learning experiment. Once you have tried, tested, and trained your experiment, you can use it to start generating predictive analysis for your actual data and expose them as web services as well. The following diagram summarizes these core concepts in a concise manner:

One of the biggest benefits of Azure Machine Learning Studio is that there are multiple algorithms that are available out of the box as follows:

• Anomaly Detection: Identifies and predicts rare or unusual data points, such as predicting a credit risk or detecting a fraudulent activity.
• Classification: Identifies what category new information belongs to. It could be a simple yes/no or true/false categorization, or a complex one such as the sentiment analysis of a social media post.
• Clustering: Separates similar data points into intuitive groups, such as predicting customer taste or the level of interest in a particular item.
• Recommendation: Predicts the likelihood that a user would prefer an item based on a previous user interaction. This is typical in online shopping portals, where additional product options are provided based on your past browsing history.
• Regression: Forecasts the future by estimating the relationship between variables, such as predicting car sales figures for the next few months.
• Statistical Functions: Mathematical computations on column values, calculating correlations, z scores, probability scores, and so on.
• Text Analytics: Works with both structured and unstructured text to detect the language of input text, creates n-gram dictionaries, and extracts the names of people, places, and organizations from unstructured text.
• Computer Vision: Image processing and image recognition tasks to detect faces, image tagging, color analysis, and so on.

As you can see, there are multiple algorithms already available; however, developers/data scientists often need to extend or modify some of the functionalities, so also possible to use R or Python script modules and Jupyter notebooks.

In terms of execution, as we mentioned previously, you can expose your machine learning workflow to external applications using REST web services. In this, there are two broader options based on the type of execution models as follows:

• Request Response Service (RRS): For stateless, synchronous execution
• Batch Execution Service (BES): Batch processing in an asynchronous way