Our services
We provide tailored support for your business decisions in various ways:
- Advisory services for specific decision challenges
- Custom decision support solutions aligned with your data and objectives
- Trade-off and impact analysis using our unique framework
We offer a full range of services, from ideation to decision, or specific steps in between, depending on your needs. We can assist using existing tools or create bespoke solutions.
With a background in aerospace engineering, software development, and expertise in uncertainty quantification and machine learning, we are equipped to address a range of decision challenges in policy, engineering, manufacturing, and healthcare, including:
- Technology evaluation
- Environmental impact assessment
- Supply chain optimization
- Renewable energy integration
- Climate risk assessment
- Lifecycle assessment
Many decision problems require a similar approach, and therefore we developed a framework that works for problems in a wide variety of industries ranging from aerospace and automotive to energy, transport, construction and governments.
Our framework
What is it?
- Define clear goals and scope.
- Formulate insightful questions to uncover underlying needs.
- Identify key stakeholders.
- Explore potential solution directions and offer initial design and planning.
- Establish a plan for the next steps.
Our approach
- Engage in meaningful dialogues with stakeholders and domain experts to gather relevant information.
- Foster open, no-commitment discussions to explore potential collaboration opportunities.
- Conduct research and analysis to deepen understanding.
Examples
- Aircraft Efficiency Enhancement: Determine key aerodynamic technologies to investigate with the goal of reducing fuel burn per flight. Specify the desired level of detail in modeling these technologies to capture their effects accurately.
- Medical Diagnosis Support: Define which symptoms and potential causes (e.g. diseases, allergens) should be correlated. Healthcare professionals observe a patient’s history of symptoms and habits and employ a probabilistic model to pinpoint the most likely cause of symptoms. This enhances diagnostic accuracy and improves treatment decisions.
- Supply Chain Optimization: Minimize pollution and waste of a production and distribution process for a given product by selecting alternative materials and manufacturing processes, and redistributing suppliers.
What is it?
- Develop a model of the problem domain, which may include a knowledge graph encoding facts, a machine learning model, or a probabilistic model.
Our approach
- Initiate with a proof of concept that captures the fundamental aspects of the decision problem, validating data, the model, and expectations.
- Iteratively enhance and expand the model to align with the requirements of all stakeholders.
- Prioritize creating holistic models that sufficiently cover the problem domain, allowing for ongoing improvement based on observed outcomes.
Examples
- Aircraft Fuel Burn Model: Develop a set of models that capture the behavior of aircraft components, and ultimately relate technology effects to the aircraft fuel burn.
Diet-Health Relationship Model: Develop a probabilistic model linking foods to irritants and irritants to symptoms. Implement an application for patients to log their daily intake and symptoms, enabling personalized health insights.
Life Cycle Analysis: Define the functional unit and perform an inventory analysis relevant to the product under consideration. Construct different models for each scenario (with different materials and/or processes) that each compute energy consumption, pollution and waste.
What is it?
- Data collection/extraction: Gather the necessary data for the decision-making process.
- Model inference: Perform inference using the established models.
- Uncertainty quantification: Quantify and manage uncertainty associated with the data and models.
- Impact assessment: Assess the impact of the data and uncertainty on decision outcomes.
Our approach
- Initiate quantification by obtaining input data for the models from various sources, such as databases, text mining, or expert input.
- Develop data-driven methods to learn and adapt the data from available observations.
- Propagate input data through the models, especially when considering uncertainty. This may involve techniques like Monte Carlo analysis or probabilistic inference to account for uncertainty.
- Continuously update and refine quantification as new data becomes available or as models improve.
Examples
Technology Impact Assessment: Assign probability distributions to the effects of different technologies and perform a Monte Carlo analysis to propagate uncertainty to estimate the impact on fuel burn.
Cause-Effect Probability Assessment: Assign probabilities to cause-effect relationships in the context of symptoms, observe actual data, and quantify the probabilities of specific irritants causing the symptoms. Update these probabilities as new data and insights emerge.
What is it?
- Perform sensitivity analysis to assess the model’s behavior and data.
- Evaluate if the results align with expectations and examine the influence of key parameters on the outcomes.
Our approach
- Conduct sensitivity analysis to gauge the model’s responsiveness to parameter variations.
- Engage in discussions with stakeholders and experts to validate and refine the results.
- Calculate and assess model performance metrics to measure the model’s accuracy and reliability.
Examples
- Parameter Sensitivity Analysis: Identify that the model is highly sensitive to engine efficiency, as expected. Measure the sensitivity of various technology effects, allowing for the attribution of differences in fuel burn to specific technologies.
- Uncertainty-Driven Insensitivity: Observe that certain cause-effect relationships exhibit extreme uncertainty and are relatively insensitive to the available observations. Exercise caution when interpreting results and do not rule out the presence of these causes.
- Model Accuracy: Validate the model’s accuracy based on historic data and determine that its recommendations align with expectations, providing confidence in the decision-making process.
What is it?
- Address decision problems, which typically involve trade-offs between alternatives or the search for the optimal solution.
- Assist in making the best possible decision, considering various scenarios and uncertainties.
Our approach
- Employ probabilistic inversion techniques to determine the best alternative among a set of candidates.
- Use optimization techniques that are robust and reliable, especially in the face of uncertainty, to design the best solution.
- Provide a complete application that can be reused for recurring decision problems, streamlining the decision-making process.
Examples
- Technology Ranking: Rank technologies from best to worst performing, considering different scenarios or aggregating results over scenarios based on the probability of each scenario.
- Health & Nutrition Application: Develop an application that informs healthcare professionals about the foods most likely to be the cause of symptoms.
- Robust Supply Chain: Determine the most robust and reliable configuration for a production process that minimizes pollution, energy consumption and waste given variability in material and energy availability, regulation and market.
How can we help you making data-driven decisions?
Have any questions? We are always open to talk about your business, and how we can assist you in making the most complex decisions.