U.S. Department of Energy

Pacific Northwest National Laboratory

Temporal Modeling for Streaming Analytics

Streaming data has a natural temporal component as data is observed and collected in a sequential manner. We are developing a framework that improves statistical modeling and predictive analytics on data streams by leveraging the temporal structure of the data and including temporal dependence in probabilistic learning models. This framework will allow researchers to overcome issues of integrating data and information sources that comes in at different temporal resolutions and irregular time scales. Methodologies, visualizations, and algorithms will be presented in an interactive environment, which provides transparency to the user and allows for informed decision-making.


Rather than developing methods to get around temporal dependence, we will leverage the temporal structure and information in data streams. This will lead to better predictions, accuracy, and model performance. These probabilistic models will execute under stringent space and time constraints of streaming analytics. Additionally, methods will allow for multiple data sources to be integrated and included in models, even in cases where data comes in different and/or irregular rates.


This research data will be useful for anyone who wants to make data-driven decisions. These methods will be general enough that they should be applicable over most domain spaces. These methods will outperform traditional statistical models and algorithms that ignore temporal structure that are currently being used on streaming data.

Domain experts and decision-makers will benefit from:

  • improved predictions
  • better forecasts
  • deeper understanding of the underlying data structure

Additionally, they will benefit from the ability to leverage multiple data sources without constraint on the format of data sources, frequency of data collection, etc.

| Pacific Northwest National Laboratory