Serial offenders by their very nature are not easily identified in terms of generic predictors for future offences? Elements such as the McDonald triad for example have been vastly criticised and even then it is not constant predictor in all serial offenders?
When performing a feature selection task such as this, it would be very helpful to perform principal component analysis (PCA). It is used to narrow down an input/problem's dimensionality by analyzing which inputs (precursors/predictors of being a serial offender in your case) offer the most information in your classification task.
With this method, you would begin with a large, labeled data set containing data from many predictors and narrow them down to the several which correlate most with your problem.
Here is a good resource to learn more about PCA: http://www.iro.umontreal.ca/~pift6080/H09/documents/papers/pca_tutorial.pdf
Good luck!
https://www.amazon.com/African-American-Serial-Killers-Neglected-Investigative/dp/1797910051/ref=sr_1_8?keywords=Janet+McClellan&qid=1551735634&s=gateway&sr=8-8 You might find the book noted in this link of interest, however, I think a variable that we might have to consider when creating profiles is that to do so 'internationally' might run afoul of national, ethnic, and historical differences between peoples.
I am a great fan of both Jung and Campbell but as such, in my study of their works, they raise something of the question at too broad a brush of environmental issues (national social constructs, ethnicity within those constructs, historical implications/restraints or traditions/conditions) that make cross-culture & cross-nationality comparisons troublesome. That does not mean impossible, it just means that ethnography considerations should be examined/acknowledged as well.
- Badges
- Science topic
- Similar topics
- Data Mining
- Classification