Resample regression
Posted: Tue Jun 09, 2015 7:54 am
I am following these steps to calculate a confidence interval:
1. Estimate the regression coefficients
2. Keeping these residuals as the original sample, generate B bootstrap samples.
Then calculate bootstrap Y∗b,i values for each observation in the bootstrap sample.
3. Regress the bootstrapped Y∗b,i values on the fixed x values to obtain estimated
bootstrap regression coefficients. Estimates are calculated by least-square regression.These
estimators can be used to construct the bootstrap standard error
and confidence intervals for the regression coefficients.
(from here:http://digitalcommons.georgiasouthern.e ... ontext=etd)
I wonder how I do step 3. How do I regress the bootstrapped values? I now have them in a series.
1. Estimate the regression coefficients
2. Keeping these residuals as the original sample, generate B bootstrap samples.
Then calculate bootstrap Y∗b,i values for each observation in the bootstrap sample.
3. Regress the bootstrapped Y∗b,i values on the fixed x values to obtain estimated
bootstrap regression coefficients. Estimates are calculated by least-square regression.These
estimators can be used to construct the bootstrap standard error
and confidence intervals for the regression coefficients.
(from here:http://digitalcommons.georgiasouthern.e ... ontext=etd)
I wonder how I do step 3. How do I regress the bootstrapped values? I now have them in a series.