stepback: Fitting a linear model by backward-stepwise regression

Description

stepback fits a linear regression model applying a backward-stepwise strategy.

Usage

stepback(y = y, d = d, alfa = 0.05, family = gaussian(), epsilon=0.00001 )

Arguments

dependent variable

data frame containing by columns the set of variables that could be in the selected model

alfa

significance level to decide if a variable stays or not in the model

family

the distribution function to be used in the glm model

epsilon

argument to pass to glm.control, convergence tolerance in the iterative process to estimate de glm model

Value

stepback returns an object of the class lm, where the model uses y as dependent variable and all the selected variables from d as independent variables.The function summary are used to obtain a summary and analysis of variance table of the results. The generic accessor functions coefficients, effects, fitted.values and residuals extract various useful features of the value returned by lm.

Details

The strategy begins analysing a model with all the variables included in d. If all variables are statistically significant (all variables have a p-value less than alfa) this model will be the result. If not, the less statistically significant variable will be removed and the model is re-calculated. The process is repeated up to find a model with all the variables statistically significant.

References

Conesa, A., Nueda M.J., Alberto Ferrer, A., Talon, T. 2005. maSigPro: a Method to Identify Significant Differential Expression Profiles in Time-Course Microarray Experiments.

Examples

Run this code


## create design matrix
Time <- rep(c(rep(c(1:3), each = 3)), 4)
Replicates <- rep(c(1:12), each = 3)
Control <- c(rep(1, 9), rep(0, 27))
Treat1 <- c(rep(0, 9), rep(1, 9), rep(0, 18))
Treat2 <- c(rep(0, 18), rep(1, 9), rep(0,9))
Treat3 <- c(rep(0, 27), rep(1, 9))
edesign <- cbind(Time, Replicates, Control, Treat1, Treat2, Treat3)
rownames(edesign) <- paste("Array", c(1:36), sep = "")
dise <- make.design.matrix(edesign)
dis <- as.data.frame(dise$dis)


## expression vector
y <- c(0.082, 0.021, 0.010, 0.113, 0.013, 0.077, 0.068, 0.042, -0.056, -0.232, -0.014, -0.040, 
-0.055, 0.150, -0.027, 0.064, -0.108, -0.220, 0.275, -0.130, 0.130, 1.018, 1.005, 0.931,
 -1.009, -1.101, -1.014, -0.045, -0.110, -0.128, -0.643, -0.785, -1.077, -1.187, -1.249, -1.463)

s.fit <- stepback(y = y, d = dis)
summary(s.fit)

Run the code above in your browser using DataLab

Last chance! 50% off unlimited learning