Matt Wand's Faculty Seminar: The Data Explosion

In his Faculty seminar on 24 Sept, Professor Matt Wand explained the joys of semiparametric regression, based on his co-authored book on the topic. As a method of fitting curves and surfaces to data, it combines the simplicity of standard regression methods (such as lines of best fit) with the flexibility of data mining methods data smoothing. The combination means that models can be designed for specific purposes, incorporating domain knowledge but still sensitive to data. Matt surveyed the wealth of data that such methods are now applied to, ranging from barcode supermarket data to microarray data in genomics.

Slides of the talk are available.