The purpose of my project was to see if it is possible to successfully use an already established formula(Rossmo's formula), which is used to predict the residences of serial murderers, rapists and arsonists based on their past crimes, for graffiti crimes.
This process involved taking a set of known graffiti locations and matching them with four known vandalswhose residences were approximately known. Using Matlab, I was able to write programs that allowed me to perform various regressions to solve for empirically determined constants and exponents in Rossmo's formula.
Theoretically, these values could then be used to try and predict the locations/residences of unidentified graffiti vandals. Data on graffiti locations and vandal residences were acquired by a combination of interviews with the Santa County Sheriff's Department, the Santa Cruz city planning department; there was also some data simulation required
After all empirically determined values were solved for to most accurately fit the crimes of all four vandals, the algorithm in general did not produce probability contours that consistently overlapped with the actual offender residences.
Overall, the algorithm has very minimal predictability value for graffiti crimes. Either a different equation needs to be utilized or mathematical modeling is not the most efficient means of locating offenders.
This project is to examine the usefulness of an algorithm, which is used to catch seriel criminals, when applied to graffiti crimes.
Science Fair Project done By Alanna K. Williams