A question on number of observations:
I am running the "npregress" command with bandwidth specification: cross validation, Discrete kernel: liracine, and continuous kernel: epanechnikov. I have a total observations of 239, but this regression takes into account of only 216 observations. As a result, i will have only 216 predicted values for the dependent variable. Could you please suggest me what could be the reason? or, how can I address this problem? Thanks in advance, Shyam
↧
Re: Nonparametric regression: Like parametric regression, but not
↧