The SCA Statistical System and B34S Econometric System provide a wide spectrum of predictive modeling, data mining, and scoring capabilities using parametric, semi-parametric, and non-parametric approaches. Predictive modeling is used widely in information technology (IT), business planning and operations, and industrial process control. In customer relationship management, for example, predictive modeling is used to identify customers that have purchased certain products that are likely to to purchase a related product or service based on demographics, past history, and personal profile data. Other applications of predictive modeling include customer retention, capacity planning, financial asset management, credit fraud detection, engineering, healthcare, meteorology and city planning.

Parametric Methods

Regression-based models

Multiple-input time series models

Multivariate time series models

Logistic, Probit, Multinomial logistic models

Semi-Parametric Methods

Multivariate adaptive regression splines (MARSpline)

General additive models (GAM)

Non-Parametric Methods

Projection pursuit regression (PPREG)

Alternating conditional expectations (ACE)

PI-Spline Models

Model optimization using boosting and bagging methods to improve accuracy and reliability

Classification and regression trees