Machine Learning Service (MLS)
Machine Learning Service (MLS) aggregates abundant distributed algorithms and various data pre-processing and machine learning algorithms for you to orchestrate training, evaluation, and prediction processes of machine learning models on a visualized operation interface. In addition, it seamlessly integrates data analysis and prediction applications to offer an easy-to-use, high performance, and scalable big data analysis platform for data mining and analysis.
Join the open beta to claima limited free trial. Learn more
Machine learning models of multi-dimensional features can be efficiently built, because the computing engine is built on Spark and the engine and machine learning algorithm architectures support scale-out. The computing engine can be decoupled and MapReduce mainstream computing engines are supported.
MLS integrates mainstream open source distributed algorithm libraries, for example, MLLib.
Open architecture enables you to customize algorithms and components.
MLS provides easy-to-use data analysis methods. The visualized workflow design interface greatly lowers skill requirements on data analysis personnel.
You can extract customer features and similarities by classification. This help you understand distribution characteristics of customers with different behavior and then make business decisions and perform service activities.
You can efficiently group customers using the aggregation (Kmeans) algorithm to obtain similarities of customer groups. For example, the aggregation results provide reference for carriers to customize package, market new devices, and recommend value-added services.
You can make predictions by learning from events in the past and summarizing their similarities and association. For example, in the banking domain, you can analyze customers' assets and relationships using regression algorithms to predict their total financial assets and obtain trustworthy consumption tags.
Seamless Connection to Spark
The computing engine of MLS instances is created based on Spark by default. As a result, MLS instances can seamlessly connect to Hadoop and Spark and use them for content analysis.
MLS provides elastic expansion capabilities towards big data. Clusters can be flexibly deployed and expanded to meet actual service requirements.