|
on Discrete Choice Models |
| By: | A. PRINZIE; D. VAN DEN POEL |
| Abstract: | Random Forests (RF) is a successful classifier exhibiting performance comparable to Adaboost, but is more robust. The exploitation of two sources of randomness, random inputs (bagging) and random features, make RF accurate classifiers in several domains. We hypothesize that methods other than classification or regression trees could also benefit from injecting randomness. This paper generalizes the RF framework to other multiclass classification algorithms like the well-established MultiNomial Logit (MNL) and Naive Bayes (NB). We propose Random MNL (RMNL) as a new bagged classifier combining a forest of MNLs estimated with randomly selected features. Analogously, we introduce Random Naive Bayes (RNB). We benchmark the predictive performance of RF, RMNL and RNB against state-of-the-art SVM classifiers. RF, RMNL and RNB outperform SVM. Moreover, generalizing RF seems promising as reflected by the improved predictive performance of RMNL. |
| Date: | 2007–06 |
| URL: | http://d.repec.org/n?u=RePEc:rug:rugwps:07/469&r=dcm |
| By: | Christian Belzil (Centre National de la Recherche Scientifique, CIRANO and IZA); Marco Leonardi (University of Milan and IZA) |
| Abstract: | We develop a non-rational expectation econometric model of sequential schooling decisions. Using unique Italian panel data in which individual differences in attitudes toward risk are measurable (with error), we investigate the effect of risk aversion on the probability of entering higher education. This allows us to characterize the subjective (as opposed to the objective) effect of higher education on marginal risk exposure. Because the measure of risk aversion (the classical Arrow-Pratt degree of absolute risk aversion) is posterior to schooling decisions, it depends on current wealth realizations and we must therefore take into account its endogeneity. We also allow risk aversion to be measured with error. After taking into account both the endogeneity of wealth and measurement error, we find that risk aversion is a key determinant (comparable to parents’ educational background) of the decisions to enter higher education. Precisely, risk aversion acts as a deterrent to higher education investment. |
| Keywords: | risk aversion, ex-ante risk, schooling, subjective beliefs, dynamic discrete choices |
| JEL: | J24 |
| Date: | 2007–08 |
| URL: | http://d.repec.org/n?u=RePEc:iza:izadps:dp2994&r=dcm |