Hi Pranav,
You should increase the number of trees. By default, it is set to 10,
which would explain why you don't reach higher precision.
Best,
Gilles
On 3 June 2014 07:32, Pranav O. Sharma emailpra...@gmail.com wrote:
Hi,
I'm trying to use
Thanks a lot. With 100 trees, I'm getting double digit precision now. Is
that the max you can get with 100 trees?
predicted by rf...
[[ 0.42 0.58]
[ 0.64 0.36]
[ 0.39 0.61]
[ 0.39 0.61]
[ 0.54 0.46]
[ 0.58 0.42]
[ 0.71 0.29]
[ 0.25 0.75]
[ 0.68 0.32]
[ 0.41 0.59]
[ 0.38 0.62]
If your trees are fully developped (which is the case by default),
then class probabilities represent
the proportion of trees predicting the given class. So yes, with 100
trees, precision is limited to 2 digits.
On 3 June 2014 08:29, Pranav O. Sharma emailpra...@gmail.com wrote:
Thanks a lot.
Hi,
I'm trying to use
http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html#sklearn.ensemble.RandomForestClassifier.predict_proba
and wondering if the predict_proba function can return probabilities with
higher precision (more than 1 digit).
Thanks a lot,
