Daniel San Martín,
José Manuel Gutiérrez,
Miguel Ángel Rodríguez,
3rd Open Call
The challenge requests an anomaly detection system to handle over 150MB of weather forecast information per minute. The forecasts provided gather information for a large set of weather variables stored in hundreds of GRIB files: a concise format, commonly used in meteorology. In the case of the weather model used by UBIMET, these GRIB files contain information for different weather variables, and are sequentially produced during the model execution.
Our proposal is to treat the data as a stream of messages, yielding a fast processing rate and using the available tools and algorithms.
The solution will implement a highly scalable anomaly detection tool that detects anomalies in near real time data streams. The tool is also applicable to many other industries, besides the current challenge.
Predictia is a spin-off company that emerged in 2008 from the data mining group at the University of Cantabria. We offer data management and mining solutions for problems that require a special handling for storage, access and visualisation. Our focus is developing solutions for two main sectors: climate and health.
As a spin-off company, we work on the interface between the academic and the business world. Since 2009 and 2010 respectively, we work continously with the Spanish Ministry of Health, and Food and Agriculture Organization of the United Nations. Our portfolio of private clients include BSH Group, EDP Energias de Portugal and Iberdrola.To boost research and development, we participate in the European 7th Framework, Horizon2020 and Copernicus programs.