Automatic Selection of Box Cox Transformation Parameter

An automatic selection of the Box Cox transformation parameter is estimated with two methods.
Guerrero's (1993) method yields a lambda which minimizes the coefficient of variation for subseries of x. For method "loglik", the value of lambda is chosen to maximize the profile log likelihood of a linear model fitted to x. For non-seasonal data, a linear time trend is fitted while for seasonal data, a linear time trend with seasonal dummy variables is used.

BoxCoxLambda(x, method = c("guerrero", "loglik"), lower = -1, upper = 2)

Arguments

x

a numeric vector or time series

method

method to be used in calculating lambda. Can be either "guerrero" (default) or "loglik".

lower

lower limit for possible lambda values, default is -1.

upper

upper limit for possible lambda values, default is 2.

Value

a number indicating the Box-Cox transformation parameter.

Note

This function was previously published as BoxCox.lambda() in the forecast package and has been integrated here without logical changes.

References

Box, G. E. P. and Cox, D. R. (1964) An analysis of transformations. JRSS B 26 211–246.

Guerrero, V.M. (1993) Time-series analysis supported by power transformations. Journal of Forecasting, 12, 37–48.

See also

BoxCox

Author

Leanne Chhay and Rob J Hyndman

Examples

lambda <- BoxCoxLambda(AirPassengers, lower=0)