NY Times reported about Mike Flowers, who used quantitative measures to find apartments or buildings that were overcrowded. The typical method was to use random checks in areas or go to areas that had some complaints about overcrowding in the apartments. The hit rate was about 10%, which means less than random and pretty much useless when the building inspectors hit random buildings. However, when his group started crunching the data, the data was in bad shape. They needed to get the data into order and then find correlates that were important.
Once they did, Instead of 10% correct rate, it jumped to nearly 70% hit rate for illegal conversions into multi-family buildings. When you calculate that with the 900,000 property lots in the city, then it is possible to determine the enormous productivity of the method.
Interestingly, they used all the data they had at the moment – whether there had been foreclosure, delay in property tax, missed payments, service cuts, fire data from 5 years to correlate.