araastat / reprtree Goto Github PK

Categorical data cutpoints based on binary expansion in getTree. Need to implement in as.tree

Labels are plotted as if plot.tree(..., type='uniform') is not in effect.

I'm attempting to use this for the random forests I've generated using the sklearn package in python, and my code runs fine in R using repl_python(), but when I get to the step of creating the tree (repr <- ReprTree(py$mod, py$data1)) it stays that it can't be run due to "Error in UseMethod("predict2"). Full error below:

Error in UseMethod("predict2") :
no applicable method for 'predict2' applied to an object of class "c('sklearn.ensemble._forest.RandomForestClassifier', 'sklearn.ensemble._forest.ForestClassifier', 'sklearn.base.ClassifierMixin', 'sklearn.ensemble._forest.BaseForest', 'sklearn.base.MultiOutputMixin', 'sklearn.ensemble._base.BaseEnsemble', 'sklearn.base.MetaEstimatorMixin', 'sklearn.base.BaseEstimator', 'python.builtin.object')"

Has anyone dealt with this before, and if so, is there a way around it?

hi, i've seen people also posting similar issue that i get when trying to use the package to plot random forest results from the caret fit. Any way around it? Here is the error I am getting:

zz <- getTree(rf.best$finalModel, k=1, labelVar=TRUE)
realtree <- reprtree:::as.tree(zz, rf.best$finalModel)
>> Error in data.frame(var = fr$var, splits = as.character(gTree[, "split point"]),  : 
  arguments imply differing number of rows: 117, 0

No other function is available to R. e.g.

reprtree::plot.getTree(mdl_rf$finalModel)
Error: 'plot.getTree' is not an exported object from 'namespace:reprtree'

I'm using Microsoft Open R 3.4.0 and RStudio 1.0.153 on OS X Sierra 10.12.6.

Thanks

Hello !

I tried to use the function reprtree:::plot.getTree(my_model) but it is not working and I have the following message:
"Error in row.names<-.data.frame(*tmp*, value = strtoi(x, 2)) :
duplicate 'row.names' are not allowed
In addition: Warning message:
non-unique values when setting 'row.names': "

I tried to display my trees with the following lines of code:
rf11_100<-randomForest(Bid~., data=df11, ntree=100, proximity=TRUE)
reprtree:::plot.getTree(rf11_100,3,labelVar=TRUE)

(3 is just to get the display of the tree number 3).

Let me know if you need anything else.

Thanks for your help !

Visign

Hello,

I used the "casualForest" function from "causalTree" package, once I want to find the best tree by using "reprtree" function, I got the following error.

I really appreciate any help.

dataset:
N observation:9000
outcome variable: categorical (0,1)
W=categorical

cf <- causalForest(formula=fmla, data=forestX, treatment=forestX$W,
split.Rule="CT", split.Honest=T, split.Bucket=F, bucketNum = 5,
bucketMax = 100, cv.option="CT", cv.Honest=T, minsize = 2L,
split.alpha = 0.5, cv.alpha = 0.5,
sample.size.total = floor(nrow(forestX) / 2), sample.size.train.frac = .5,
mtry = ceiling(ncol(forestX)/3), nodesize = 100, num.trees=10,
ncolx=length(predictorsNames),ncov_sample=4)

ReprTree(cf,forestX,metric="d2")
[1] "Constructing distance matrix..."
Error in UseMethod("predict2") :
no applicable method for 'predict2' applied to an object of class "causalForest"

Reproducible example:

library(reprtree)
library(randomForest)
data(spam, package="kernlab")

rf <- randomForest(type ~. , data = spam, ntree = 10)
as.tree(getTree(rf, 1, labelVar = T), rf)

Produces:

Error in match.fun(FUN) : object 'factor.repr' not found

current default is left splits

The comment above is the error message I receive in R v 3.5.1 when I attempt to install the reprtree package using the code below:

`options(repos='http://cran.rstudio.org')
have.packages <- installed.packages()
cran.packages <- c('devtools','plotrix','randomForest','tree')
to.install <- setdiff(cran.packages, have.packages[,1])
if(length(to.install)>0) install.packages(to.install)

library(devtools)
if(!('reprtree' %in% installed.packages())){
install_github('araastat','reprtree')
}
for(p in c(cran.packages, 'reprtree')) eval(substitute(library(pkg), list(pkg=p)))`

Any thoughts on how to fix this?

Reproducible example:

library(reprtree)
library(randomForest)
data(spam, package="kernlab")

rf <- randomForest(type ~. , data = spam, ntree = 1)
reptr <- ReprTree(rf, spam, metric = "d2")

Produces:

[1] "Constructing distance matrix..."
Error in x[l, ] : subscript out of bounds

Solution: something like:
if ntrees = 1, then representative tree is that tree

Great package, thx
Could we add node IDs to the plot ?

Thanks!
Markus