Many univariate simple quasi poisson regressions {Rfast}

Many univariate simple poisson regressions

Description

It performs very many univariate simple poisson regressions.

Usage

quasi.poisson_only(x, y, tol = 1e-09, maxiters = 100)

Arguments

x

A matrix with the data, where the rows denote the samples (and the two groups) and the columns are the variables. Currently only continuous variables are allowed.

y

The dependent variable; a numerical vector with many discrete values (count data).

maxiters

The maximum number of iterations after which the Newton-Raphson algorithm is terminated.

tol

The tolerance value to terminate the Newton-Raphson algorithm.

Details

The function is written in C++ and this is why it is very fast. It can accept thousands of predictor variables. It is usefull for univariate screening. We provide no p-value correction (such as fdr or q-values); this is up to the user.

Value

A matrix with the deviance and the estimated phi parameter (dispersion parameter) of each simple poisson regression model for each predictor variable.

Author(s)

Manos Papadakis <[e-mail]> and Stefanos Fafalios <[e-mail]>

R implementation and documentation: Michail Tsagris <[e-mail]>, Manos Papadakis <[e-mail]> and Stefanos Fafalios <[e-mail]>.

References

McCullagh, Peter, and John A. Nelder. Generalized linear models. CRC press, USA, 2nd edition, 1989.

See Also

poisson_only univglms, logistic_only, allbetas, regression

Examples

## 200 variables, hence 200 univariate regressions are to be fitted
x <- matrix( rnorm(100 * 200), ncol = 200 )
y <- rpois(100, 10)
poisson_only(x, y)
b1 <- poisson_only(x, y) 
b2 <- quasi.poisson_only(x, y) 

b1<-b2<-x<-y<-NULL