Graph of scikit-learn ExtraTreeClassifier and RandomForestClassifier -

i trying make graphs illustrate difference between randomforestclassifier , extratreeclassifier in scikit-learn. think might have figured out unsure. here code fit , graph iris dataset both:

import numpy np sklearn.datasets import load_iris sklearn.externals.six import stringio  sklearn import tree import pydot  iris = load_iris()  x = iris.data y = iris.target  clf = tree.extratreeclassifier()  clf = clf.fit(iris.data, iris.target)   dot_data = stringio()  tree.export_graphviz(clf, out_file=dot_data)  graph = pydot.graph_from_dot_data(dot_data.getvalue())  file_name = "et_iris.pdf" graph.write_pdf(file_name)   clf = tree.decisiontreeclassifier()  clf = clf.fit(iris.data, iris.target)   dot_data = stringio()  tree.export_graphviz(clf, out_file=dot_data)  graph = pydot.graph_from_dot_data(dot_data.getvalue())  file_name = "rdf_iris.pdf" graph.write_pdf(file_name)  

the graphs produces seem correct, et graph "bushier" decision tree graph.

am correct decisiontreeclassifier same single tree in randomforestclassifier , extratreeclassifier same single tree in extratreeclassifier?

is there way trees in actual rdf or et classifier? tried using .estimators_ in forests not have export method seems.

export_graphviz not method, function. none of trees "has" it. can use estimators_. right extratreeclassifier being single tree in extratreesclassifier , decisiontreeclassifier being single tree in randomforestclassifier. however, doesn't cover it, because:

• randomforestclassifier bootstraps data set separately each tree, extratreesclassifier not bootstrap (by default).

• max_features=n_features single trees default, features can used in every split.