Neural nets and the logic to estimate the costs of related matters

At some point we will be able to analyze data on a moderate numbers off cases and predict the costs of similar cases.  An inspiration for this forecast is a computer system that predicts which death-row inmates are most likely to be executed, as described in Scientific Am., Vol. 299, Sept. 2008 at 36.  Software analyzed data on about 1,000 death-row prisoners, including their sex, age, race, schooling and whether they were ultimately executed.  When given similar information about 300 more prisoners, the logic developed from the first set of data correctly predicted the outcome for 92 percent of those cases. 

 

Someday, when a group of law departments contribute sufficient data on certain matters, someone will use similar software to predict the likely cost of such matters (See my post

pril 7, 2006 

analytic software for invoices; and 

005

on artificial intelligence software.).  It will take 20-30 legal departments who each contribute relevant data on 20 to 30 matters – such as recently concluded employment discrimination or patent suits, to reach a comparable level of insight (See my post of Sept. 22, 2006: entrepreneurial ideas on this blog.).