Salford Customer Wins 1st Place in Prestigious Competition Using Salford Predictive Modeler Software Suite
Using the Salford Predictive ModelerÒ Software Suite, DataLab has been awarded first place honors in the 2016 Direct Marketing Association Analytics Challenge.
San Diego, CA: The annual Direct Marketing Association (DMA) Analytic Challenge winners were announced last month in Los Angeles, CA, naming DataLab USA as one of the First Place co-winners. This year’s competition was sponsored by the DMA and EY and challenged teams to predict customer expected value and drive personalized customer interactions. Using the Salford Predictive Modeler (SPM) Software suite v8.0, the DataLab team built one of the most accurate models possible to predict customer expected value.
The DataLab team experimented with two approaches for this year’s challenge; they used financial calculations as well as using a direct model approach and found that an ensemble of the two approaches garnered the best results.
When choosing the tools in their kit, the DataLab team focused primarily on predictive power, speed, and the ability to tune model parameters. Using the SPM TreeNetÒ Software in their DataLab Predictive Modeling Toolkit, the team was able to clinch this win. “TreeNet has demonstrated a significant speed advantage versus other implementations,” said Aaron G. Davis, Vice President of Analytics at DataLab USA, a Salford Systems client. “Salford has also provided us with the ability to tune the model across many advanced dimensions such as the ability to penalize interactions and/or variables, which ultimately allows us to develop a more discriminant and better generalizing model.”
Salford Systems congratulates DataLab USA on their success and strives to build data mining tools with robust product performance, including unparalleled speed, accuracy and scalability.
In addition to powerful proprietary automation and modeling capabilities, the Salford Predictive Modeling Suite includes the company's four data mining products:
- CART® (Classification and Regression Trees): Noted for its modeling accuracy, performance, visualization, feature set automation and ease of use, CART quickly reveals important data patterns and relationships that could remain hidden using other analytical tools.
- TreeNet®: TreeNet is the proprietary technology underlying major recent advances in fraud detection, targeted marketing, and risk modeling. Its easy to use and enables users to create super-accurate, targeted marketing models and identify ultra-high lift segments with little analyst supervision. TreeNet has the advantage of being able to work effectively with dirty and erroneous data, separating problematic from reliable information. It is also responsible for the majority of the data mining competitive wins Salford has secured in the last decade.
- MARS® (Multivariate Adaptive Regression Splines): Ideal for users who prefer results in a form similar to traditional regression while capturing essential nonlinearities and interactions. Its flexibility permits MARS to trace out non-linear patterns detected in the data. It can predict continuous numeric outcomes as well as probability models for yes/no outcomes.
- RandomForests®: Best suited for analyzing complex data structures embedded in small-to-moderate data sets for deep understanding. It uses the power of multiple alternative analyses, randomization strategies and ensemble learning to produce accurate models, insightful variable importance rankings and accurate reporting.
About Salford Systems: Founded in 1983, Salford Systems specializes in providing new-generation data mining and predictive modeling software and consulting services for industries such as banking, insurance, healthcare, pharmaceutical, telecommunications, transportation, manufacturing, retail and catalog sales, and education. The company's software is currently installed in over 3,500 sites worldwide, including 300 major universities. Salford Systems is headquartered in San Diego, CA. For more information, visitwww.salford-systems.com or telephone (619) 543-8880.