Extendable System for Multicriterial Outlier Detection
The article is devoted to developing a convenient extendable software system for outliers detection in data. The developed information system is based on the use of statistical analysis and machine learning methods to find suspicious values in data and the use of Web technologies and micro-service architecture to implement the user interface and system extensibility. As a development result, a software system was implemented that can analyze multidimensional numerical data and find outliers in them using a set of customizable analysis methods with the opportunity to vote algorithms. New algorithms could be easily added to the system as microservices interacting with the parent Web-service. End users can access the system through the Web application using any Web browser. The developed system can be used in data analysis and to process experimental results, which can potentially contain errors. This delivers the necessary degree of automation for an expert analyzing the data correctness.