PATENT LITIGATION RISK-SCORING MODEL

We develop a model to predict the probability of a patent being involved in litigation in the near future. Litigation is defined as a legal filing, either offensive or defensive in nature. The model will be used by insurance companies to better evaluate potential clients for intellectual property insurance with a coverage time of two years. The data for the model are from a random sample of all patent case filings from 1994 through 2000. We added a control group of patents not involved in litigation that are similar to the litigated patents. We gather potentially predictive variables about the patent and its owner to build the data set for regression. A logistic regression model is built in SAS using stepwise regression.