Nonparametric test for checking lack of fit of the quantite regression model under random censoring

2008 ◽  
Vol 36 (2) ◽  
pp. 321-336 ◽  
Author(s):  
Lan Wang
Keyword(s):  
Regression Model ◽  
Nonparametric Test ◽  
Random Censoring ◽  
Lack Of Fit

scholarly journals A note non Nonparametric Regression Modeling using a Density Function

2020 ◽  
Vol 7 (2) ◽  
pp. 993-1000
Author(s):  
Jakperik Dioggban
Keyword(s):  
Regression Model ◽  
Density Function ◽  
Nonparametric Regression ◽  
Model Simulation ◽  
Smoothing Splines ◽  
Bias Reduction ◽  
Simulation Studies ◽  
Lack Of Fit ◽  
Classical Regression ◽  
Better Than

The nonparametric regression offers alternative to classical regression analysis when the data are not well behaved or when the classical regression model shows significant lack of fit. In recent years, It has been applied using Kernel estimators and the smoothing splines, but these methods wields some bias of estimation. In this study, a semi-parametric multiplicative bias reduction density function was used to develop a non parametric regression model. Simulation studies conducted showed that the proposed estimator performs better than both the Kernel and the smoothing splines estimators especially with large samples

Indeterminacy

2017 ◽  
Author(s):  
Russell Cheng
Keyword(s):  
Regression Model ◽  
Null Hypothesis ◽  
Alternative Hypothesis ◽  
Hypothesis Test ◽  
Null Distribution ◽  
Elementary Approach ◽  
Test Statistic ◽  
Exponential Regression ◽  
Lack Of Fit ◽  
Exponential Regression Model

This chapter discusses models like the exponential regression model y = a[1− exp(− bx)] where if a = 0 then b is an indeterminate, non-identifiable parameter, as it vanishes from the model. The hypothesis test that H0 : a = 0 versus H1 : a ≠ 0 is then non-standard. The well-known Davies test is explained. This uses a portmanteau test statistic T that is a functional of Sn(b), L< b< U, where Sn(b) is a regular test statistic of the null hypothesis a = 0 versus the alternative a ≠ 0 with b fixed. The null distribution of T is not usually easy to obtain. One can instead just test if a = 0 using a GoF test or a lack-of-fit test with an alternative hypothesis not specified. In the exponential regression example, this means simply testing if the observations are solely pure error. This elementary approach is compared with the Davies approach.

Sign in / Sign up

Export Citation Format

Share Document