Getting the most out of your Data 

Interactive Data Exploration

In many application areas, large amounts of data are available from physical observations and/or computer models. It can be a tough task to separate important information from the overall data volume, and to turn the raw data into knowledge. At Kalkulo, we have developed several applications that incorporate interactive data exploration through smart co-visualization of data from different sources, customized filtering, and statistical analysis.

Such functionality is part of our software tools for the oil and gas sector and are also centerpieces of the software products MetOcean and Torch. MetOcean combines different sources for meteorological, climate, and environmental data in order to analyse the situation in ice-covered ocean areas. Torch is a software tool for condition monitoring, performance assessment, and business decisions for wind farms. These products make use of Kalkulo's platform for data integration and decision support, DIDSy.


Data-Driven Forecasting through
simulation and machine learning

While data exploration and data sifting represent first steps in handling the streams of data encountered in many applications, the next step is often to use these data for predictions. This can be done in different ways. For systems governed by the laws of physics, observations can be input to numerical models trying to mimic Nature. These models can then be used to estimate where the underlying physical system is headed. The weather forecast or the predicted flow of fluids in a pipeline are examples of combining observations with such dynamic simulation models, typically based on solving differential equations posed on a discretized geometry.

An alternative to encoding physics to mathematical equations and computer simulations, is to build the model directly on the available data in a probabilistic manner. Such data-driven models can also be derived for problems that can not be described by the classical continuum mechanical approach. Machine learning, which refers to a multitude of methods spanning from simple regression models to deep convolving neural networks, is the most important approach to data-driven forecasting.

Kalkulo has expertise in process simulation and machine learning across several application areas. As needed, we also collaborate closely with relevant research groups in these areas, in particular the Department of Numerical Analysis and Scientific Computing at Simula Research Laboratory and Department of Machine Intelligence at the Simula Metropolitan Center for Digital Engineering. One particularly interesting line of work is to develop hybrid approaches where simulations and machine learning strengthen each other.



Complementing our expertise on data exploration and data-driven forecasting, Kalkulo has long experience in designing and building software solutions that interface seamlessly into existing and planned digital workflows. The key to digitalization is to see how the various digital puzzle pieces of your organization and business are connected, and how they, step by step, can brought to the level of autonomous interplay. At Kalkulo, we approach such transformative processes in a down-to-earth and pragmatic way, making sure that every step provides value to our customer.


Do you need help in turning your data into value? Please feel free to contact us to discuss your digital challenges.

> Numerics