Applications of the new kind of modeling

Focus, insight, and trust are the three distinguishing features of the technology that make DataModeler shine when dealing with real world data-driven problem solving.

Focus is the added value coming from the ability to automatically spotlight important variables only and use this information in data analysis and predictive modeling. DataModeler creates “white-box” models — which contrasts with the opaque black-box variety generated by conventional data mining methods. Predictive models are explicit mathematical expressions designed to have minimal complexity and automatically focus on the driving variables only is undeniably important in real-life applications.

The game changer in creating insight is that it finally became easy to analyze many variables together rather than one-at-a-time, to discover rather than impose the nature of variable relationships and do it on real data with missing and unbalanced records, correlated and spurious variables, often with only partial a priori knowledge.

Trust is a result of the salient technology in DataModeler to quantify prediction uncertainty. Whatever is the empirical model - it cannot be trusted unless its prediction comes with a reliable confidence measure.

All applications in process and research analytics which generate data are pre-destined to benefit greatly from DataModeler's unique features. Examples of industries and disciplines where success is declared are below (the list is not mutually exclusive):

  • Chemical Industry — inferential sensors, empirical emulators and product design,
  • Manufacturing — inferential sensors, production process optimization,
  • Energy — predictive modeling, energy system management, emission minimization,
  • Epidemiology — non-linear predictive modeling and variable identification in epidemiological studies, infectious disease modeling,
  • Statistics — model structure identification, meta-variable discovery for statistically validated non-linear modeling,
  • Simulation and Simulation-based Optimization — high-fidelity empirical emulators for computation intensive fundamental models and black-box simulators,
  • Research, Development, Discovery Informatics — design space exploration, active design of experiments,
  • System Identification — model structure discovery.