Random Forest Regressor using a custom objective/ loss function (Python/ Sklearn)

python-3.x scikit-learn random-forest statsmodels poisson
vishmay picture vishmay · Mar 26, 2018 · Viewed 7k times · Source

I want to build a Random Forest Regressor to model count data (Poisson distribution). The default 'mse' loss function is not suited to this problem. Is there a way to define a custom loss function and pass it to the random forest regressor in Python (Sklearn, etc..)?

Is there any implementation to fit count data in Python in any packages?

Answer

Marcus V. picture Marcus V. · Mar 26, 2018

In sklearn this is currently not supported. See discussion in the corresponding issue here, or this for another class, where they discuss reasons for that a bit more in detail (mainly the large computational overhead for calling a Python function).

So it could be done as discussed within the issues, by forking sklearn, implementing the cost function in Cython and then adding it to the list of available 'criterion'.

Related questions

Python RandomForest - Unknown label Error

I have trouble using RandomForest fit function This is my training set P1 Tp1 IrrPOA Gz Drz2 0 0.0 7.7 0.0 -1.4 -0.3 1 0.0 7.7 0.0 -1.4 -0.3 2 ... ... ... ... ... 3 49.4 7.5 0.0 -1.4 -0.3 4 47.4 7.5 0.0 -1.4 -0.3 ... (10k rows) I want to predict P1 thanks to all the other variables using sklearn.…

Read More
"ValueError: max_features must be in (0, n_features] " in scikit when using random forest

I have a dataset of 20 features and 840 rows. I have already optimized the classifier (random forest). My parameters are n_estimators=100 and max_features=5. I want to do a classification for each feature. I mean with each of the features …

Read More
ValueError: Expected 2D array, got 1D array instead:

While practicing Simple Linear Regression Model I got this error, I think there is something wrong with my data set. Here is my data set: Here is independent variable X: Here is dependent variable Y: Here is X_train Here …

Read More