Import Safety by Coglianese Cary;Finkel Adam M.;Zaring David;

Author:Coglianese, Cary;Finkel, Adam M.;Zaring, David;
Language: eng
Format: epub
Publisher: University of Pennsylvania Press

is replaced by

Each predictor has its own inductively produced functional relationship with the response, which are then combined in an additive fashion.

Although the main goal in profiling is to forecast accurately, many machine/statistical learning procedures can provide measures of the forecasting contribution of each predictor. The basic idea is to compute how much less accurate a forecast is when any given predictor is not allowed to contribute to the forecasting exercise. Then, predictors may be ordered by their forecasting “importance.”

It is also possible to show the inductively generated functional form linking each predictor to the response. For example, Figure 2 shows how the age of individuals on parole or probation is related to the likelihood that they will commit a homicide or attempted homicide, with other predictors held constant. The technical details need not concern us here. When the parolee or probationer is under 20 years of age, the risks are the highest. The risk drops precipitously up to about age 30, after which age seems to have virtually no relationship to the outcome. Thus, although a difference in age between an individual of 18 and 25 can be very important, the difference in age between an individual of 28 and 35 hardly matters at all.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.