Comments for Jared Lander

Comment on The FBI Needs Your Help by Jose Galofre Manero-SPAIN_europe

Jose Galofre Manero-SPAIN_europe — Wed, 09 Nov 2011 00:21:05 +0000

People:
Look the 2nd note,last line,said:
“O-W-m-4 H8L XORLX”
Can you see what letter are repetead?
IF L=M and X= Variations of the letter C (MC, C, CK) then
“OWN FOR I AM MCCORMICK”
This is a SIGNATURE and Mccormick´s LAST WILL.
This is the key master
Bye.

Comment on Supreme Court Justice Rules for NY Pizza by Pete(I Love Pizza)Anderson

Pete(I Love Pizza)Anderson — Tue, 08 Nov 2011 18:14:41 +0000

Great ruling hope all the best in the future Las Vegas Pizza

Comment on The FBI Needs Your Help by Marie

Marie — Fri, 01 Apr 2011 23:59:10 +0000

Ricky McCormick is a genius in his own right. He thinks in layers and writes his codes in the same manner. On his sheet you will find clues, symbols and codes. He may have only placed a few clues on his sheets of paper. One clue leads to another clue and another code. Think of it as a treasure hunt. I don’t know much about puzzles, I look at things a different way. I read pictures of people and use psychometry to figure things out. This is what I feel and see when I look at this. Latitude and Longitude. Ricky thinks in numbers and layers of complexity.. If you take the latitude and longitude of St. Louis Missouri you get this: Latitude 38.646991 and Longitude -90.224967 If you take the last two digits off each coordinate you get this: 38.646974 latitude and -90.224975 longitude. It’s the same place but look around to see what’s in the area and you this business called Piece of PI on Cozens Ave and Francis Street in St. Louis MO. Friends? Address? Code? If it’s a code then the next clue is that he created a code based on PI and that means random numbers. He may also have created a PI based code language. http://investigacio.wordpress.com/category/pi-language/
If you look in the top right of his paper it says PI. I also saw what looked like longitude or latitude marks and math type symbols. If someone didn’t kill him for what he saw or assumed he saw they may have killed him to gain rights to his codes if he showed or shared them with anyone. I’m thinking he might be using some kind of latitude longitude grid on top of a Peace of PI symbol that is written in a circle like a compass or wheel then using this to point to numbers that are a PI language. He thinks in layers so it’s got to be something like that. Good luck. This is the best I can come up with.

Comment on Top Pizzerias by Blog Shout-Outs by Jared

Jared — Wed, 02 Mar 2011 18:08:14 +0000

The division is there to compare a clustering of k groups versus a clustering of k+1 groups, so I don’t think the division destroys the effect. It’s sort of like a likelihood ratio test. And then the (N-k-1) is a standardizing constant.

I believe, W_{k+1} should be larger than W_k, so the ratio will be less than one which is indeed small, and multiplying by (N-k-1) helps see if that change is really significant.

I do get what you’re saying, but I think the ratio is typically less than 1 (which I think it is), so you have to think about it a little differently.

Then again, I could be off on the ratio, so check into that before taking what I say at face value.

Comment on Top Pizzerias by Blog Shout-Outs by Thanh

Thanh — Wed, 02 Mar 2011 08:13:22 +0000

thanks for your discussion, and for the link

I have just found this problem of the Hartigan Rule interesting to me. I tested the Rule with a sample data set (where the number of objects is about 10000), and noticed that the number of natural clusters suggested by Hartigan Rule is really high.

I think, the Within Sum of Squares (often call W_k) is important. But, in the Hartigan Rule, the division (W_k / W_{k + 1}) tends to destroy the effect of the Within Sum of Squares. The ratio (W_k / W_{k + 1}) will not be very small. So, when multiplied by (N – K – 1), the suggested number of clusters is very high. What do you think ?

Comment on Top Pizzerias by Blog Shout-Outs by Jared

Jared — Wed, 02 Mar 2011 06:05:22 +0000

That is a fair point. But it is there to see if the change in Within Sum of Squares is significant or just due to large data. See this great PowerPoint from David Madigan: http://www.stat.columbia.edu/~madigan/DM08/descriptive.ppt.pdf

Comment on Top Pizzerias by Blog Shout-Outs by Thanh

Thanh — Tue, 01 Mar 2011 05:16:47 +0000

I am a little confused.
In the computation of the Hartigan numbers, there is N (that is, the number of entities in the data set) with multiplication.
So, if N is very large (for example, N = 1000), the number of clusters will be very high. What do you think about this case ?

Comment on Predictive Analytics Wrap Up (Update: Harlan’s Writeup) by zerodtkjoe

zerodtkjoe — Wed, 20 Oct 2010 10:40:11 +0000

Thanks for the info

Comment on Coefficient Plot by livelybrowsers

livelybrowsers — Tue, 19 Oct 2010 18:09:00 +0000

Thanks for good stuff

Comment on Coefficient Plot by RedMango

RedMango — Mon, 18 Oct 2010 12:34:24 +0000

Very nice post!