eco.lmtree: Fitting Multiple Linear Regression models by stepwise AIC selection and Multiple Classification and Regression Trees via party

Description

This program fits for each dependent variable, a Multiple Linear Regression model calling the function step for choosing the best model by AIC criterion, or a Multiple Classification and Regression Trees model, using the package party. The summary of the model returns information about the significance of the models, F-statistics and degrees of freedom, when is fitted a "mlm"; otherwise, when the model fitted is a "mctree", the summary returns the plots of those trees with significant splits.

Usage

eco.lmtree(df1, df2, analysis = c("mlm", "mctree"), mod.class = "+",
  fact = NULL, ...)

Arguments

df1

Data frame with dependent variables as columns.

df2

Data frame with independent variables as columns.

analysis

Class of analysis to perform. "mlm" for multiple linear regression analysis, or "mctree" for a multiple classification tree analysis.

mod.class

"+" for additive model, "*" for model with interaction, in both cases, these models will include all terms in the dependent data frame. If other model than these two is desired, it could be specified as a string with the names of those columns of the independent variable that should be used as terms. This string corresponds to the right side "x" of a formula y ~ x (see examples).

fact

Optional factor for estimating the frequencies of individuals from different levels in each node, when the analysis performed is "mctree".

...

Further arguments passed to lm or ctree

Value

When the analysis selected is "mlm", the output object has three main slots:

> MLM: the results of the model

> SUMMARY.MLM the summary for each variable returned by the lm function

> ANOVA.MLM with the ANOVAs results.

When the analysis selected is "mctree", the output object has also three main slots:

> TREES: Trees returned by the multiple ctree analysis.

> PREDICTIONS: Predictions of the analysis.

> FREQUENCIES: Number of individuals predicted in each node.

ACCESS TO THE SLOTS The content of the slots can be accessed with the corresponding accessors, using the generic notation of EcoGenetics (<ecoslot.> + <name of the slot> + <name of the object>). See help("EcoGenetics accessors") and the Examples section below

References

Hothorn T., K. Hornik, and A. Zeileis. 2006. Unbiased Recursive Partitioning: A Conditional Inference Framework. Journal of Computational and Graphical Statistics, 15: 651-674.

Examples

Run this code

# NOT RUN {
# }
# NOT RUN {
data(eco.test)

# mlm additive model
mod <- eco.lmtree(df1 = eco3[["P"]], df2 = eco3[["E"]], 
analysis = "mlm")                                 
mod
summary(mod)

# mctree additive model
mod <- eco.lmtree(df1 = eco3[["P"]], df2 = eco3[["E"]], 
analysis = "mctree", fact = eco3[["S"]]$pop) 


#-----------------------
# ACCESSORS USE EXAMPLE
#-----------------------

# the slots are accessed with the generic format 
# (ecoslot. + name of the slot + name of the object). 
# See help("EcoGenetics accessors")

summary(mod)

ecoslot.FREQUENCIES(mod)        # slot FREQUENCIES

# frequency table with counts of individuals in populations x terminal nodes
tabfreq <- do.call(cbind, ecoslot.FREQUENCIES(mod))
namestab <- lapply(ecoslot.FREQUENCIES(mod), ncol)
namestab <- lapply(namestab, rep)
namestab <- rep(names(namestab), namestab)
colnames(tabfreq) <- namestab
tabfreq

# mlm custom model
mymod <- "E1+E2*E3"
mod <- eco.lmtree(df1 = eco[["P"]], df2 = eco[["E"]], 
analysis = "mlm", mod.class = mymod)            
summary(mod)

# mctree custom model
mod <- eco.lmtree(df1 = eco[["P"]], df2 = eco[["E"]], 
analysis = "mctree", mod.class = mymod, fact = eco[["S"]]$pop)   
 
                      
summary(mod)

# }
# NOT RUN {
# }

Run the code above in your browser using DataCamp Workspace