fraudulent data frame has 42 rows and 12 columns.
127 claims arising from automobile accidents in 1989 in Massachusetts (USA). Each claim was classified as either fraudulent or legitimate by consensus among four independent claims adjusters who examined each case file thoroughly. An exploratory analysis by Derrig and Weisberg (1993) identified 10 binary indicators, each of which denotes the presence or absence of a potential fraud characteristic in the claim situation. They fall into three broad groups relating to “Accident” (AC1, AC9 and AC16), “Claimant” (CL7 and CL11), and “Injury” (IJ2, IJ3, IJ4, IJ6 and IJ12).
This data frame contains the following columns:
the number of frauds detected;
the total number of automobile insurance claims;
potential fraud characteristics pertaining to “Accident”. The presence of the fraud characteristic is indicated by a 1, the absence is indicated by a 0.
potential fraud characteristics pertaining to “Claimer”. The presence of the fraud characteristic is indicated by a 1, the absence is indicated by a 0.
potential fraud characteristics pertaining to “Injury”. The presence of the fraud characteristic is indicated by a 1, the absence is indicated by a 0.
The data were supplied by Dr. Richard Derrig of the Automobile Insurers Bureau of Massachusetts.
Mehta, C. R., Patel, N. T. and Senchaudhuri, P. (2000) Efficient Monte Carlo methods for conditional logistic regression. J. Amer. Statist. Ass., 95, 99–108.
Derrig, R. A. and Weisberg, H. I. (1993). Quantitative methods for detecting fraudulent automobile bodily injury claims. Manuscript.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.