ARBOLES DE CLASIFICACION CON FACTOR DE PONDERACION APLICADO AL ESTUDIO DEL CONSUMO DE TABACO EN JOVENES DE LA REGION METROPOLITANA, CHILE

Abstract

The objective of this paper is to describe the profile of the Region Metropolitana’s (RM) students that have smoked cigarettes or that have consumed some kind or tobacco during the last month. Classification Trees with Weight factor are used, using the database of the Tobacco Consumption Inquiry by Youngs 2000 (EMTAJOVEN, March 2000, OMS, MINSAL). The sample consisted in 3150 students between 12 and 15 years old. 26 categorical variables, related with personal characteristics and the consumption of tobacco were measured and the weight factor. Equal a priori probabilities were specified and the incorrect classification costs were 1, 1.5, 2, 2.5 ad 3.0. For validating, a test sampling was used and the specificity, sensitivity and correct classification rates were tested on the construction and validation of 13 different trees the expanded samples. The final
classification tree determined 8 variables for describing the different student groups with a specificity rate of 80% and a sensitivity of 89%. The important variables are related with personal characteristics, place where he/she smokes, intention of smoking, exposition to tobacco’s smoke, the believe that to smoke light cigarettes is less dangerous and that smoker adolescents have more or less friends