Predictive analytics and forecasting is a frequent data innovation challenge. Typical tasks are the detection a trends and anomalies in historical data, the prediction of future events, etc. It's applications are manifold and appearing in multiple domains, e.g., predictive maintenance for industrial machines, predicting traffic flows in mobility, the detection of anomalies while monitoring health parameters, etc. Data is typically coming from sensors, measuring a multitude of parameters, and coming continuously at a high speed.
In order to make data innovation tangible, several demonstrators will be developed in selected domains. These will focus on key data innovation challenges (e.g., entity profiling, predictive analytics), and illustrated by scenarios and use cases in relevant application domains. For each demonstrator a generic reference architecture will be designed, with references to relevant technological platforms, toolkits, and libraries. Furthermore, guidelines, best practices and lessons learnt regarding data characteristics, algorithms, validation, etc. will be distilled. These generic demonstrator concepts can subsequently be tailored towards a company context and validated on company-specific data and technologies, creating specific data processing workflows.
Entity profiling is the attempt to gather essential characteristics of an entity to characterise its typical behaviour. Entities can be people, organisations, events, etc. Examples of entity profiling can be found in different domains, such as lifestyle profiling based on personal and social information, competitive intelligence based on publicly-available legal data, or the profiling of websites in terms of security and performance.