Consider the scenario where a practical problem is tackled utilizing the method of least squares. Upon each iteration, an estimation of the parameter $\theta$ is derived via $\hat{\theta} = (X^\top X)^{-1}X^\top Y$, where $X$ and $Y$ denote arrays of $x_i$ and $y_i$ for $i$ ranging from 1 to $n$. However, it is pertinent to acknowledge the result may differ in each estimation. Let's say N values of $\hat{\theta}$ are derived from N estimations, each of which is an outcome obtained using arrays of $x$ and $y$. Is there a methodology to assess the confidence associated with $\hat{\theta}$ procured from each least squares estimation?
4
-
$\begingroup$ Confidence intervals for the coefficients? $\endgroup$– DaveNov 2 at 2:31
-
$\begingroup$ @Dave could you please be more specific? $\endgroup$– yangtzechNov 2 at 14:05
-
$\begingroup$ I'd encourage you to read about regression confidence intervals. Perhaps these lecture slides from a Georgia Tech course could be a starting point. $\endgroup$– DaveNov 2 at 14:11
-
$\begingroup$ @Dave Sorry for the late reply. Great attitude to your kind help, and I'll try to look into it first. $\endgroup$– yangtzechNov 8 at 1:59
Add a comment
|