"Data mining" involves identifying various features of the data recorded for specific types of crimes to determine such factors as crime location, victim characteristics, offender characteristics, time, and weapons used. "Predictive analytics" involves the identification of risk factors associated with particular types of crime so as to predict where and when specific types of victims and offenders are likely to become involved in certain types of offenses. The principles of "data mining" and "predictive analytics" are illustrated in an analysis of drug-related homicides in Richmond, VA. Using predictive analytics, reliable models for drug-related homicides were developed. The models revealed a relatively complex relationship between suspect attributes and victim lifestyle. This analysis afforded the opportunity for law enforcement agencies to identify specific factors that could be addressed through targeted prevention or selective enforcement strategies. One benefit of the characterization of victim risk factors is the identification of variables that have contributed to police officer assaults and on-duty officer deaths. Such findings can be incorporated in officer training to improve safety techniques. Further, predictive analytics can be used to identify geographic areas where various types of offenses have occurred in the past. This information facilitates the strategic and cost-effective deployment of police resources to areas where various crime types are likely to occur. Strategies for officer action can be designed to counter those factors that have been linked to criminal behaviors and victim behaviors.