G - Physics – 06 – F
Patent
G - Physics
06
F
G06F 17/30 (2006.01) G06F 17/60 (2000.01)
Patent
CA 2320605
The present invention relates to a method and apparatus for local Bayesian classification and regression. In a first step, a joint estimate is made of the density in the covariate and response variables constituting multivariate normal distributions. Subsequently responses are sampled from the estimate and finally an arbitrary transformation is performed on latent parameters of a k-dimensional distribution, where each dimension is defined on a real line (i.e. - ~ to + ~) and where k is an integer greater than or equal to 2. The invention offers extremely flexible ways of modeling data and is capable of dealing with a highly varied customer base that does not follow a global trend. It is unaffected by outliers, such as errors in the data or small groups of customers displaying unusual behavior. It can also deal elegantly with missing values in the data. An important technical contribution of the invention is that it allows inferences to be reversed from a given behavior to demographic descriptions of customers that perform that behavior. In other words it can provide a previously unobtainable executive summary describing the factors that drive customer behavior.
Ncr Corporation
Smart & Biggar
LandOfFree
Classifying data in a database does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Classifying data in a database, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Classifying data in a database will most certainly appreciate the feedback.
Profile ID: LFCA-PAI-O-1968989