Work with FICO experts to tailor solutions, services and training to meet your needs
FICO Optimization Consulting Services
The FICO® Optimization Consulting Services approach uses and conveys methodologies, software and best practices developed and refined through years of research and across hundreds of client engagements. This breadth of experience means FICO offers organizations the highest confidence in achieving success when applying optimization and prescriptive analytics to improve business-critical decisions. In particular, FICO has more experience
than any vendor in developing optimization solutions for credit decisions — which are often complex and require a high degree of accuracy.
A typical FICO Optimization Consulting Services development project will usually focus on developing a new set of improved decisions that will deliver a strong return-on-investment (ROI). To achieve this, the FICO Professional
Services team works collaboratively with the client team to design and develop the required Optimization, Decision & Predictive models. These are then set up in FICO’s Xpress Optimization software to run and assess a range of simulation and optimization scenarios using different business goals and constraints. The chosen scenario can then be built out or deployed via a number of options. This may be as an application, a process or as decision
logic (usually as rules or in a decision tree format) that can be easily deployed within or integrated with production systems.
FICO increasingly uses these projects to also support the client team in adopting pre-configured solutions, optimization software tools and development methodologies, so they can independently obtain ongoing improvements, both from the initial project, and also more widely across a range of decision areas.
FICO Analytic Consulting Services
FICO's pedigree in predictive and prescriptive modelling is second to none. Started in 1956, FICO has pioneered the development of Credit Scoring around the world. FICO’s analytic innovations include:
We continually pioneer the development of new predictive and prescriptive tools and techniques to meet today’s business needs.
We regularly undertaken Analytic Consulting Services and model developments as part of client engagements to develop Optimization models, services and solutions. These often are focused on business problems where we combine the knowledge of business domain, analytic and optimization experts. Typical business problems we regularly support include: Marketing, Fraud, Collections & Recoveries, Account & Customer Management, Origination, Credit Bureau, Regulatory / Capital Management, Supply Chain, Econometric Forecasting etc.
The types of Analytics we typically develop include:
Decision (or Prescriptive) Models: These describe the relationship between all the elements of a decision—the known data, predictions, the decision and the forecast results of the decision. They are increasingly used with optimization solvers to run different scenarios, which maximise certain outcomes while minimising or constraining others. Their main purpose and application is to identify improvements in decision strategies, but they’re also used for scenario analysis and stress testing. FICO is a pioneer in this area having been developing different types of decision and prescriptive models for well over 15 years.
Predictive Models, Machine Learning & AI Development
Predictive models analyse past performance to assess how likely a customer is to exhibit a specific future behaviour, often rank ordering customers by their likelihood or probability of taking a particular action. Some of the predictive modelling techniques and approaches used by FICO include:
- Ensemble Models: Refers to techniques where the predictions of a group of base models are combined to generate more accurate composite predictions. Ensemble modeling usually involves two main activities: 1) Constructing an ensemble of base learners from training data; 2) Combining predictions of ensemble members into a composite prediction
- FICO Scorecard Technology: These are FICO’s proprietary predictive modelling techniques. They are unique in their ability to account for business rules, legal and operational constraints, and biased or missing data. As a result, FICO can generate scores that are not only highly predictive, but are also very interpretable and palatable to the business user. The patented algorithms use mathematical solver technology to optimise score weights for binary, multiple goal and continuous outcomes.
- Neural Networks: This is an approach that transforms a set of inputs, via linked, directed and weighted interconnections, into a set of predictive model outputs. FICO pioneered the use of neural networks, alongside patented customer profiling techniques, to create powerful predictions in areas such as fraud detection.
- Time-to-Event Models: A type of discrete time hazard model developed by FICO which looks to identify the probability of an event occurring over different time periods, in effect predicting when events are most likely to occur.
- Action Effect/Uplift Models: This is a type of causal model that infers from available data the effects of different potential actions on outcomes to help identify which would likely provide the best outcome. FICO has developed a range of techniques mainly focused on marketing and optimization projects.
- Demand Elasticity / Take up Models: Often used in Pricing Optimization, these focus on how the quantity of a product or service demanded changes with its price (or other factors, such as Lending Amount).
- Economic Impact Models: These models provide predictions that are adjusted by changes in macro-economic and market conditions, and thus help better understand both current and anticipated future conditions.
Descriptive Models: These models quantify relationships in data in a way that is often used to classify customers or entities into groups based on their similarities, while maximising segment differences. They often identify many different relationships between customers or products, but do not usually rank-order customers by their likelihood or probability of taking a particular action.
- FICO Analytic Consulting
- Understanding Predictive Analytics - Learn about the different types of predictive analytics and how they improve business decisions. Information level: Basic.
- Analytic Technologies - Explore more than 20 technologies used to develop predictive analytics. Information level: Advanced.