As a company, our goal is to help Scientists, Engineers, Managers, Manufacturers, Government Agencies, and Individuals understand their project outcomes — through mechanistic modeling, data analysis, visualization, and interdisciplinary communications.
What do these people have in common?
What do a Manager of a utility company trying to predict and reduce future repair costs, a Lawyer representing clients whose property was damaged due to plumbing that failed far before its expected lifetime, and a Clothing Manufacturer wanting to provide clothing that fits the widest variety of different body types have in common with a Fire Department Chief altering fire station staffing to meet new regulatory performance guidelines?
They all need these tools:
- Measurement: what data can and should you collect? How much information do you already have? How accurately can you determine the real value in the world of the quantities of interest to you? How much money will different investigation strategies cost and which ones will inform you the most?
- Dynamics: The world is constantly changing, during and between measurements. Change occurs for multiple reasons. Causes that occur at one time can have affects that are only observable at other later times.
- Physical Laws: Forces, conservation laws, energy balances, chemical reactions, and many other scientific principles can be used to build models of causal processes, not just statistical correlations. Sometimes these are exact and other times they are approximate.
- Variability and Uncertainty: Conditions vary from place to place, and from time to time, as does our ability to measure them. Unless we acknowledge uncertainty and build it into our analysis, we can't know whether a difference between prediction and reality is meaningful or just an expected consequence of unrelated noise.
- Comparison: Naive comparisons of alternatives under uncertain conditions can lead to misleading conclusions. Contractor A bid less than contractor B so A was given the project. What would have happened if we had chosen contractor B? Could it ultimately have led to fewer cost overruns and lower final costs? What evidence do we have for that conclusion vs alternative conclusions?
- Optimization: Optimization of a problem down to the single best value of all possible choices is rarely realistic. Most problems involve many noisy evaluations of real world quantities. Searching through a complex landscape of options for solutions that are within some reasonably good set is generally more realistic, and requires a model of what we mean by "reasonably good". In real world examples, defining what is to be optimized is potentially the most important part of any optimization analysis.
- Interdisciplinary connections: No one can be expert in all aspects of science, mathematics, statistics, business, economics, communication, politics, and all the other myriad skills necessary to solve complex problems in the real world. We specialize in extracting knowledge from domain experts, connecting that knowledge to other domains where we have expertise, building mathematical models to describe the process, and visualizing data to understand the results. The output of our investigations can feed back to experts in other domains and help them refine their own understanding.