Investigative Data Mining for Security and Criminal Detection

8.13 Enhancing the Data

In addition to transactional variables listed above, demographic information is associated to each customer account in a typical data warehousing technique. Demographic information is associated based on a zip code or a geo code matched against a shopper's physical address. This type of demographic data has typically been used for marketing purposes to segment potential prospects on the basis of their household and neighborhood features. In this situation, however, it is used for the profiling of perpetrators and the detection of criminal activity. Other data points can be included in the development of a fraud-detection model, incorporating additional clickstream information, which, again, is commonly used for marketing purposes but not fraud detection. This clickstream data can be created from such Internet mechanisms as Web bugs, cookies, and Web forms. The demographic data includes some of the following information; the following is a partial listing of over 150 lifestyle variables:

• GEOCODE: A location code down to 200 household level

• MSA_CODE: Metropolitan statistical area

• LATITUDE: Location field

• LONGITUDE: Location field

• ZIP4: Zip plus four

• ACORN: A Classification of Residential Neighborhoods code

• POP_CY: 2000 population

• AVGHHSIZE: Average household size

• P_WHITE_CY: Percent White population

• P_AGE25_44: Percent 25-44 population

• MED_AGE_CY: Median age — total population

• P_COLL_GRD: Percent college graduates

• P_MANAGMNT: Percent in managerial positions

• P_POVERTY: Percent in poverty level

• P_URBAN: Percent urban residences

• P_SINGLE: Percent single

• P_WOM_LABF: Percent of women in labor force

• P_BLT_1980: Percent of buildings built prior to 1980

• MEDHOMEVAL: Median home value

• MED_RENT: Median rent

• P_INC_50: Percent of income $50,000-$75,000

• MED_INC: Median income amount

• MEDDISPINC: Median disposable income

• MEDNETWRTH: Median net worth

These types of demographics can be purchased from a number of companies, including Acxiom, CACI, ChoicePoint, Equifax, Experian, Trans Union, and others, as we discussed in Chapter 2. They can be purchased on a per-record basis, via disk, CD, tape, or in real time through secured networks or via the Web, directly from the companies or through third-party integrators. These demographics can reveal very interesting lifestyle information about on-line shoppers, including those likely to carry out fraud.