Azure Machine Learning ("AML") is a Web-based computer learning service that Microsoft has launched on its public cloud azure, a branch of AI that uses algorithms to make computers recognize a large number of mobile datasets. This approach is able to predict future events and behaviors through historical data, which is significantly better than traditional forms of business intelligence. Microsoft's goal is to streamline the process of using machine learning so that developers, business analysts, and data scientists can apply it extensively and easily. The purpose of this service is to "combine machine learning dynamics with the simplicity of cloud computing". AML is currently serving on Microsoft's Global Azure cloud service platform, where users can request a free trial through the site: https://studio.azureml.net/.
After logging into the trial account, you can see the following interface:
For first-time users, you can start a built-in example by selecting "experiments" on the left-hand menu and selecting New in the lower-left corner and selecting Add a "experiments Tutorial" after the popup menu. This is a model based on existing data including age, education level, marital status, occupation, current income, etc. to predict whether the income of any class of people can exceed 50k. By clicking Next, users can easily learn how to import data, how to preprocess the data, how to separate data for training models and validate models, how to choose an algorithm to train the model, and how to evaluate the effectiveness of the model. The whole process does not require programming, it is completely easy to do by dragging and configuring. Enables users to quickly get started with AML usage, thus putting more effort into understanding data and algorithms, and the tools themselves will not bring you any additional learning costs.
At the same time, users can also use https://azure.microsoft.com/en-us/documentation/articles/machine-learning-import-data/ Learn all About Microsoft Azure machine learning. Here are some explanations for some of the most common concerns that help you get to know AML quickly and quickly.
- How data is imported and the type of data.
The data that you want to use to train and validate the model needs to be imported into studio in AML. The currently supported data import methods are as follows:
• Local File Upload
Azure BLOB Storage, table
Azure SQL Database
Hadoop using HiveQL
a web URL using HTTP
a data Feed provider (OData)
The following data types are supported:
CSV files, including. CSV and. Nh.csv;
TSV files, including. TSV and. NH.TSV;
Hadoop Hive Table
SQL database table
svmlight Data (svmlight) (Detailed description See link: http://svmlight.joachims.org/)
attribute Relation File Format (ARFF) data (. ARFF) (Detailed description See link: http://weka.wikispaces.com/ARFF)
zip file (. Zip)
• R object or workspace file (. RData)
2. Built-in algorithms
In summary, Microsoft Azure Machine learning has built up more than 20 algorithms based on supervised learning and unsupervised learning, such as classification, regression, clustering, and more, detailed algorithm descriptions see links: https://msdn.microsoft.com/en-us/library/azure/ Dn905812.aspx. I will also be in the back of the post in succession to introduce to you. In addition to the algorithms, AML integrates packages that 400+ multiple R languages.
The choice of the algorithm, whether for beginners or experienced data scientists, in fact, is a very cost-minded thing. Microsoft also provides a lot of information to help you decide which algorithms to choose. Here are a few very useful links (in English):
Microsoft Azure Machine learning algorithm Cheat sheet-https://azure.microsoft.com/en-us/documentation/articles/ machine-learning-algorithm-cheat-sheet/
Choosing a Learning algorithm in Azure machine learning-http://blogs.technet.com/b/machinelearning/archive/2015/05/ 20/choosing-a-learning-algorithm-in-azure-ml.aspx
Choosing a machine learning classifier-http://blog.echen.me/2011/04/27/choosing-a-machine-learning-classifier/
Choosing the right estimator-http://scikit-learn.org/stable/tutorial/machine_learning_map/
3. Built-in application module
In order to facilitate Microsoft Azure machine learning to make it easy for more people to get started and use, AML native built up a lot of business scenarios for raw data and machine learning modules and APIs. Users can use them directly, or make a small amount of changes for their own use. The main business scenarios include but are not limited to the following (still increasing). Beginners can first understand and master the use of machine learning from these existing modules. These already built models can be found in "Gallery" from the top menu of the login homepage.
• Text analysis;
• Customer churn forecasts;
• Referral system;
• Predictive maintenance;
• Fraud monitoring;
4. How to charge
AML is provided as a cloud service through Web Access. Both free and standard service delivery methods are currently available. The standard level is billed according to the length of use, for reference: http://azure.microsoft.com/en-us/pricing/details/machine-learning/
This article only provides a guided overview of Microsoft's Azure machine Learning service, helping you to have a preliminary understanding of the service. There is also a lot of content, including how to build a good machine learning model after the release, etc. will be in the later blog to introduce you in detail.
Microsoft Learning Azure Machine learning Getting Started overview