28.07.2014 Views

Quantile/expectile regression, and extreme data analysis

Quantile/expectile regression, and extreme data analysis

Quantile/expectile regression, and extreme data analysis

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

LMS quantile <strong>regression</strong><br />

Third Method: The Yeo-Johnson transformation† I<br />

Yeo <strong>and</strong> Johnson (2000) introduce a new power transformation which is<br />

well defined on the whole real line, <strong>and</strong> potentially useful for improving<br />

normality:<br />

ψ(λ, y) =<br />

⎧<br />

⎪⎨<br />

⎪⎩<br />

λ = 1 = the identity transformation.<br />

(y + 1) λ − 1<br />

(y ≥ 0, λ ≠ 0),<br />

λ<br />

log(y + 1) (y ≥ 0, λ = 0),<br />

− (−y + 1)2−λ − 1<br />

(y < 0, λ ≠ 2),<br />

2 − λ<br />

− log(−y + 1) (y < 0, λ = 2).<br />

The Yeo-Johnson transformation is equivalent to the generalized Box-Cox<br />

transformation for y > −1 where the shift constant 1 is included.<br />

© T. W. Yee (University of Auckl<strong>and</strong>) <strong>Quantile</strong>/<strong>expectile</strong> <strong>regression</strong>, <strong>and</strong> <strong>extreme</strong> <strong>data</strong> <strong>analysis</strong> 18 July 2012 @ Cagliari 14/101/ 101

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!