An R-squared measure of goodness of fit for some common nonlinear regression models

Article Abstract:

R-squared, or R(supra 2), is a commonly used goodness-of-fit measure for the standard linear regression model. However, a summary statistics of this familiar coefficient of determination have been constructed for other regression, nonlinear models, including logit, probit, Poisson, geometric, gamma and exponential. This R-squared is defined as the proportionate reduction in uncertainty, as measured by Kullback-Leibler divergence, due to the inclusion of regressors. It can also be viewed as the fraction of uncertainty explained by the fitted model.

Analysis, Linear models (Statistics), Goodness-of-fit tests

---

Regression-based cointegration estimators with applications

Article Abstract:

A generalized two-sided estimator is developed to determine asymptotically efficient estimates of cointegrating relationships and for single equation estimation. The cointegration equation is augmented with leads and lags of short-run variables. A non-linear single equation least squares estimator is also used for weak exogeneity. Applications include modeling stock prices and money demand.

Models, Estimation theory, Asymptotic efficiencies (Statistics)

---

Similar abstracts:

• Abstracts: Time variation of second moments from a noise trader/infection model
• Abstracts: An investigation of Ricardian equivalence in a common trends model
• Abstracts: Time variation of second moments from a noise trader/infection model. part 2 Capacity utilization and market power
• Abstracts: Using theory measurement: an analysis of the cyclical behavior of home production. Semiparametric censored regression models
• Abstracts: Tarnished Hopes. Off to Athens with Hopes for an Agreement. Hong Kong Hopes This Rat Will Run and Run

This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.