What inspired this wish list request?
Data mining "bump hunting" algorithm that does not require an empirical model. Rather, it 'patiently' / incrementally searches hyper-boxes of input factors that maximize the response.
What is the improvement you would like to see?
From R:
library<SPAN class="">(</SPAN>prim<SPAN class="">)</SPAN>
library<SPAN class="">(</SPAN>MASS<SPAN class="">)</SPAN>
data<SPAN class="">(</SPAN>Boston<SPAN class="">)</SPAN>
x <SPAN class="">\</SPAN><SPAN class=""><-</SPAN> Boston<SPAN class="">\</SPAN><SPAN class="">[</SPAN><SPAN class="">,</SPAN><SPAN class="">5</SPAN><SPAN class="">:</SPAN><SPAN class="">6</SPAN><SPAN class="">\</SPAN><SPAN class="">]</SPAN>
y <SPAN class="">\</SPAN><SPAN class=""><-</SPAN> Boston<SPAN class="">\</SPAN><SPAN class="">[</SPAN><SPAN class="">,</SPAN><SPAN class="">1</SPAN><SPAN class="">\</SPAN><SPAN class="">]</SPAN>
boston.prim <SPAN class="">\</SPAN><SPAN class=""><-</SPAN> prim.box<SPAN class="">(</SPAN>x<SPAN class="">=</SPAN>x<SPAN class="">,</SPAN> y<SPAN class="">=</SPAN>y<SPAN class="">,</SPAN> threshold.type<SPAN class="">=</SPAN><SPAN class="">1</SPAN><SPAN class="">)</SPAN>
Why is this idea important?
