Re: Nearest Neighbour / Ternary Search Tree
- From: maryjones11289@xxxxxxxxxxx
- Date: 12 Jan 2006 11:31:13 -0800
Larry/Tony,
Just to let you know, I have implimented the methods you both outlined.
Although I have some more testing ahead, with my initial tests I can
confirm that using Byte Arrays has produced significant speed
improvements.
With the dataset I am testing on, the entries in the array are not
distributed evenly. What I mean by this is that there are no entries
with a 1's count of 0 to 17 or 193 to 225, but a huge amount at around
the 112 point. If I graphed these totals, it would turn out to be a
bell shape. I only mention this as Larry indicated that I would
probably only need to compare about 20% of the entires by eliminating
those with a 1's count of +-22. This would definitely have been the
case if the entries were evenly distributed, but unfortunately for me,
they are not.
However, the code is in place, and I am limiting the compares to those
+- 22. I will continue to test, but can say I'm very happy with the
results I'm seeing currently.
Thanks again to you both and to the others that posted comments, I
appreciate it.
MJ
.
- Follow-Ups:
- Re: Nearest Neighbour / Ternary Search Tree
- From: Bob O`Bob
- Re: Nearest Neighbour / Ternary Search Tree
- References:
- Nearest Neighbour / Ternary Search Tree
- From: maryjones11289
- Re: Nearest Neighbour / Ternary Search Tree
- From: Tony Proctor
- Re: Nearest Neighbour / Ternary Search Tree
- From: maryjones11289
- Re: Nearest Neighbour / Ternary Search Tree
- From: Larry Serflaten
- Re: Nearest Neighbour / Ternary Search Tree
- From: maryjones11289
- Nearest Neighbour / Ternary Search Tree
- Prev by Date: Re: Any way to check if .mdb field exists without looping or error checking?
- Next by Date: Re: Need Help with DLL Variable Access
- Previous by thread: Re: Nearest Neighbour / Ternary Search Tree
- Next by thread: Re: Nearest Neighbour / Ternary Search Tree
- Index(es):
Relevant Pages
|