We use a combination of traditional and modern predictive techniques to derive the best predictive models. We work with complex customer data that include demographic, behavioral, and transaction elements. Our predictive modeling approaches have included Logistic and Linear Regression, CHAID techniques, Classification and Regression Trees (CART), Tobit Models, Markov Chain, Survival Modeling, and Time-Series Analysis.

Analytex has performed several types of analytic studies. Examples include Clustering and Customer Segmentation, Customer Lifetime Value Analysis, Break-Even and ROI Analyses, Customer Profitability models, Collaborative Filtering, Market Basket Analysis, and Transaction Data Analytics.

Analytex uses a proprietary methodology for constrained optimization. The methodology reduces a non-linear problem to a linear optimization problem. This technique has been employed by Analytex in solving marketing optimization problems that involve multiple objectives with several constraints.

Our data mining is especially geared to explore large amount of data (usually transaction data) to detect patterns and systematic relationships between variables. Data preparation and cleaning is the first important step in the data mining process to avoid "garbage-in-garbage-out." Data reduction techniques such as clustering, principal component analysis and feature selection are employed. We use special tools designed to handle large data with thousands of input variables and millions of records.