\
  The most prestigious law school admissions discussion board in the world.
BackRefresh Options Favorite

rach, add latex math support so that we can make cool graphics & talk about math

replace something like {\displaystyle\min_{\beta\in\mathbb{R...
Mildly autistic death wish
  10/16/18
what’s the expression in the imgur pic?
cheese-eating henna main people
  10/16/18
It's the objective function for the group lasso in the Lagra...
Mildly autistic death wish
  10/16/18
Huh. Yeah I see the Lagrangian set up. My econometrics is pr...
cheese-eating henna main people
  10/16/18
Yeah. It's a technique for selecting features and regularizi...
Mildly autistic death wish
  10/16/18
OK I think I follow. Yeah not too familiar with this area, h...
cheese-eating henna main people
  10/16/18
Model selection always occurs after feature extraction. Basi...
Mildly autistic death wish
  10/16/18
I wish the real world was amenable to even the simplest cool...
coiffed obsidian indian lodge
  10/16/18
The trick is to lessen the extent to which the data reflects...
Mildly autistic death wish
  10/17/18
Lmao
coiffed obsidian indian lodge
  10/17/18
statistics = data molestation & coercion. it's literally...
Mildly autistic death wish
  10/17/18
Thanks will do dad
coiffed obsidian indian lodge
  10/17/18
Any suggestions for simple predictive modeling besides regre...
coiffed obsidian indian lodge
  10/17/18
What are you trying to do? machine learning and, more broadl...
Mildly autistic death wish
  10/17/18
TYVM FRIEND. I will see if this is CR for my application. ...
coiffed obsidian indian lodge
  10/17/18
cr if we have emojis we should have LaTeX, RACH YOUR PEOPLE ...
cheese-eating henna main people
  10/16/18
it's so easy to add! https://jsbin.com/gofujereki/edit?ht...
Mildly autistic death wish
  10/16/18
Don't fall for this trap rach, this autistic freak will sue ...
Garnet overrated tanning salon
  10/16/18
it's a javascript library
Mildly autistic death wish
  10/16/18
...
Mildly autistic death wish
  10/16/18
...
Mildly autistic death wish
  10/17/18
Taylor do you do ML in industry?
pontificating messiness
  10/17/18
Not in a few years.
Mildly autistic death wish
  10/17/18
are you a developer? If so, do you work at a TTT or somewher...
pontificating messiness
  10/17/18
...
coiffed obsidian indian lodge
  10/17/18
I'm a shitlawyer
Mildly autistic death wish
  10/17/18
NO do not listen to TSINAH this is stupidp
Heady coldplay fan ladyboy
  10/17/18
...
Mildly autistic death wish
  10/17/18


Poast new message in this thread



Reply Favorite

Date: October 16th, 2018 4:47 AM
Author: Mildly autistic death wish

replace something like {\displaystyle\min_{\beta\in\mathbb{R}^{p}}\left\{\left\|y-\sum_{j=1}^{J}X_{j}\beta_{j}\right\|_{2}^{2}+\lambda\sum_{j=1}^{J}\|\beta_{j}\|_{K_{j}}\right\},\qquad\|z\|_{K_{j}}=(z^{t}K_{j}z)^{1/2}} with https://i.imgur.com/qJhJote.png

It's easy to do: Just add <script type="text/javascript" async src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/latest.js?config=TeX-MML-AM_CHTML" async></script> to the html template. That will let you use \( to enter inline latex mode and \) to exit inline latex mode or $$ and $$ to enter and exit equation mode (centered equation as its own line)

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033177)



Reply Favorite

Date: October 16th, 2018 5:22 AM
Author: cheese-eating henna main people

what’s the expression in the imgur pic?

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033222)



Reply Favorite

Date: October 16th, 2018 5:28 AM
Author: Mildly autistic death wish

It's the objective function for the group lasso in the Lagrangian form. Lasso = least absolute shrinkage and selection operator and the group lasso is an extension that allows predefined groups of covariates to collectively be included in or excluded from a model. See http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.79.2062&rep=rep1&type=pdf

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033224)



Reply Favorite

Date: October 16th, 2018 5:36 AM
Author: cheese-eating henna main people

Huh. Yeah I see the Lagrangian set up. My econometrics is pretty strong but I haven’t come across this LASSO technique — I either missed it, or it hasn’t come across to econometrics. Data mining technique?

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033229)



Reply Favorite

Date: October 16th, 2018 6:27 AM
Author: Mildly autistic death wish

Yeah. It's a technique for selecting features and regularizing regression models. The standard lasso selects a subset of the given covariates for use in the model, thereby limiting overfitting. The group lasso extends this idea by allowing groups of predefined covariates to be selected into or out of the model together. It's good in settings where categorical variables are encoded as a hierarchical collection of binary covariates, which is typically an instance in which random forests, for example, will underperform due to computational complexity & bias toward features with more levels. However, the lasso is by nature tailored to linear models, which isn't a constraint for random forests as decision trees are nonlinear.

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033309)



Reply Favorite

Date: October 16th, 2018 6:46 AM
Author: cheese-eating henna main people

OK I think I follow. Yeah not too familiar with this area, have never done much big data, and data mining is generally frowned on in more orthodox econometrics as letting the data determine the model, rather than letting the theory determine the model and then testing the model against the data. But I know a lot of people are getting into big data techniques and it’s something I should get up to speed on before too long.

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033323)



Reply Favorite

Date: October 16th, 2018 7:59 AM
Author: Mildly autistic death wish

Model selection always occurs after feature extraction. Basically, you look at the data and try to figure out the important variables via domain experts, dimensionality reduction techniques, heuristics, etc. Then, once you have an idea of how the independent variables affect the dependent variables, you choose a modeling approach that makes sense given the context. The hardest part is fitting (but not overfitting) the model to the data while preserving prediction accuracy.

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033435)



Reply Favorite

Date: October 16th, 2018 7:15 AM
Author: coiffed obsidian indian lodge

I wish the real world was amenable to even the simplest cool as fuck stats theories

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033356)



Reply Favorite

Date: October 17th, 2018 9:10 AM
Author: Mildly autistic death wish

The trick is to lessen the extent to which the data reflects reality via regularization or other techniques that limit overfitting

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37041067)



Reply Favorite

Date: October 17th, 2018 9:37 AM
Author: coiffed obsidian indian lodge

Lmao

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37041183)



Reply Favorite

Date: October 17th, 2018 9:39 AM
Author: Mildly autistic death wish

statistics = data molestation & coercion. it's literally rape.

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37041194)



Reply Favorite

Date: October 17th, 2018 9:41 AM
Author: coiffed obsidian indian lodge

Thanks will do dad

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37041210)



Reply Favorite

Date: October 17th, 2018 9:41 AM
Author: coiffed obsidian indian lodge

Any suggestions for simple predictive modeling besides regresssttttion?

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37041214)



Reply Favorite

Date: October 17th, 2018 9:53 AM
Author: Mildly autistic death wish

What are you trying to do? machine learning and, more broadly, artificial intelligence provide solutions to many different types of problems. Regression is just a technique for assigning a continuous value to the output variable. Classification assigns a class label to the output variable. Most regression analysis relies on linear models. If you're looking to get away from logistic or linear regression and have nonlinear data, look into random forests. They're based on decision trees (which are nonlinear models) and they're extremely good at both classification and regression, especially when combined with pruning and bagging & boosting techniques.

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37041273)



Reply Favorite

Date: October 17th, 2018 11:31 AM
Author: coiffed obsidian indian lodge

TYVM FRIEND. I will see if this is CR for my application.

How are you applying this to LAW?

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37041814)



Reply Favorite

Date: October 16th, 2018 5:03 AM
Author: cheese-eating henna main people

cr if we have emojis we should have LaTeX, RACH YOUR PEOPLE DEMAND IT

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033196)



Reply Favorite

Date: October 16th, 2018 5:17 AM
Author: Mildly autistic death wish

it's so easy to add!

https://jsbin.com/gofujereki/edit?html,output

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033215)



Reply Favorite

Date: October 16th, 2018 6:46 AM
Author: Garnet overrated tanning salon

Don't fall for this trap rach, this autistic freak will sue you for 10 BITCOINS for not properly attributing the Javascript snippet!

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033324)



Reply Favorite

Date: October 16th, 2018 8:00 AM
Author: Mildly autistic death wish

it's a javascript library

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37033437)



Reply Favorite

Date: October 16th, 2018 11:06 AM
Author: Mildly autistic death wish



(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37034328)



Reply Favorite

Date: October 17th, 2018 5:08 AM
Author: Mildly autistic death wish



(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37040755)



Reply Favorite

Date: October 17th, 2018 11:58 AM
Author: pontificating messiness

Taylor do you do ML in industry?

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37042000)



Reply Favorite

Date: October 17th, 2018 11:59 AM
Author: Mildly autistic death wish

Not in a few years.

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37042008)



Reply Favorite

Date: October 17th, 2018 12:05 PM
Author: pontificating messiness

are you a developer? If so, do you work at a TTT or somewhere where you have to study for like 6 months before interviewing

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37042034)



Reply Favorite

Date: October 17th, 2018 1:33 PM
Author: coiffed obsidian indian lodge



(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37042711)



Reply Favorite

Date: October 17th, 2018 4:02 PM
Author: Mildly autistic death wish

I'm a shitlawyer

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37043821)



Reply Favorite

Date: October 17th, 2018 12:02 PM
Author: Heady coldplay fan ladyboy

NO do not listen to TSINAH this is stupidp

(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37042022)



Reply Favorite

Date: October 17th, 2018 4:02 PM
Author: Mildly autistic death wish



(http://www.autoadmit.com/thread.php?thread_id=4107787&forum_id=2#37043823)