[ml] A challenge...

[Glen Jarvis]

This is by no means as cool as the Kaggle HIV challenge, or any of the other
challenges that you've done. Buuut, I did make a little challenge (mostly
for myself, but I'll share if you wish to take it on).

I took the CS229 lecture notes for a single variable input (Living area),
figured out a rough fit to the line and then did some 'fudging' to a new
function (I used the 'fudging' to see how close I can get to 'known'
parameters). I then added some 'noise' to the expected output (Thanks, for
the suggestion on NumPy Mike!)

So, a data file is on the Wiki Page under a new section called "Supplemental
Files" (I had to save it in Excel to be able to upload it to the wiki). If
you can figure out the exact parameters that I used to generate the data
(before adding the noise), I'll buy you a beverage. I know this isn't hard,
but it was so fun for me to generate (as I learned a lot in doing so).

I also am trying to use the algorithm listed on page 7 of the notes to see
if I can get this to converge to the parameters that I used to generate the
data. And, I'm interested in comparing other people's approaches/solutions
to this.

This also gives me a more basic example to work on before we get to the more
difficult problem sets.  Baby steps.. I need baby steps....

:)

Cheers,


Glen
-- 
Whatever you can do or imagine, begin it;
boldness has beauty, magic, and power in it.

-- Goethe
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.noisebridge.net/pipermail/ml/attachments/20100908/42a3ea6b/attachment.html>