[R-sig-Geo] finding significant local peaks by locfit package

[tkobayas at iu.edu]

HI,

I am still learning how a nonparametric regression works to find out
statistically significant local peaks on the fitted curve. My data comes from
Baltimore census tracts. The tracts have the latitude and longitude values.

lfknots showed that the fitted values, the local points, and the log likelihood
statistics. I assume the log likelihood statistics is very meaningful in this
sense. but I am not sure if there is any criteria to pick statistically
significant ones out of all sets of points evalueated by locfit.

I apprecaite if someone could help me out.

Thank you very much.

Taka


Indiana University, Dept. Geography