Download presentation

Presentation is loading. Please wait.

Published byJaden Sallis Modified over 2 years ago

1
SUG London 2007 Least Angle Regression Translating the S-Plus/R Least Angle Regression package to Mata Adrian Mander MRC-Human Nutrition Research Unit, Cambridge

2
SUG London 2007 Least Angle Regression Outline LARS package Lasso (the constrained OLS) Forward Stagewise regression Least Angle Regression Translating Hastie & Efron’s code from R to Mata The lars Stata command

3
SUG London 2007 Least Angle Regression Lasso Let y be the dependent variable and x j be the m covariates The usual linear predictor Want to minimise the squared differences N.B. Ridge regression does constraint on L2 norm Subject to this constraint, large t gives OLS solution

4
SUG London 2007 Least Angle Regression Lasso graphically The constraints can be seen below. One property of this constraint is that there will be coefficients =0 for a subset of variables

5
SUG London 2007 Least Angle Regression Ridge Regression The constraints can be seen below. The coefficients are shrunk but does not have the property of parsimony

6
SUG London 2007 Least Angle Regression Forward Stagewise Using constraints The function of current correlations is Move the mean in the direction of the greatest correlation for some small ε FORWARD STEPWISE is greedy and selects

7
SUG London 2007 Least Angle Regression Least Angle Regression The LARS (S suggesting LaSso and Stagewise) Starts like classic Forward Selection Find predictor x j1 most correlated with the current residual Make a step (epsilon) large enough until another predictor x j2 has as much correlation with the current residual LARS – now step in the direction equiangular between two predictors until x j3 earns its way into the “correlated set”

8
SUG London 2007 Least Angle Regression Least Angle Regression Geometrically Two covariates x 1 and x 2 and the space L(x 1,x 2 ) that is spanned by them μ0μ0 μ1μ1 x1x1 x2x2 x2x2 y1y1 y2y2 y 2 is the projection of y onto L(x 1,x 2 ) Start at μ 0 =0

9
SUG London 2007 Least Angle Regression Continued… The current correlations only depend on the projection of y on L(x 1,x 2 ) I.e. y 2

10
SUG London 2007 Least Angle Regression Programming similarities The code comparing Splus to Mata looks incredibly similar

11
SUG London 2007 Least Angle Regression Programming similarities There are some differences though Array of arrays… beta[[k]] = array Indexing on the left hand side… beta[positive] = beta0 Being able to “join” null matrices. Row and column vectors are not very strict in Splus. Being able to use the minus sign in indexing beta[-positive] “Local”-ness of mata functions within mata functions? Local is from the first call of Mata Not the easiest language to debug when you don’t know what you are doing (thanks to statalist/Kit to push start me).

12
SUG London 2007 Least Angle Regression Stata command LARS is very simple to use lars y, a(lar) lars y, a(lasso) lars y, a(stagewise) Not everything in the Splus package is implemented because I didn’t have all the data required to test all the code

13
SUG London 2007 Least Angle Regression Stata command

14
SUG London 2007 Least Angle Regression Graph output

15
SUG London 2007 Least Angle Regression Conclusions Mata could be a little easier to use Translating Splus code is pretty simple Least Angle Regression/Lasso/Forward Stagewise are all very attractive algorithms and certainly an improvement over Stepwise.

Similar presentations

OK

WEEK 8 SYSTEMS OF EQUATIONS DETERMINANTS AND CRAMER’S RULE.

WEEK 8 SYSTEMS OF EQUATIONS DETERMINANTS AND CRAMER’S RULE.

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google

Performance related pay ppt online Ppt on automobile related topics to economics Ppt on natural and artificial satellites around earth Ppt on basic accounting concepts Ppt on history of australia timeline Ppt on construction industry in india Ppt on amway business plan Ppt on natural resources water Ppt on satellite orbital speed Ppt on industrial revolution