MondrianForestClassifier object has no attribute partial_fit

As per the guide (https://scikit-garden.github.io/api/), partial_fit method is available for MondrianForestClassifier

MondrianForestClassifier.partial_fit(X, y, classes=None)

In google colab, I did the following,

!pip install scikit-garden
from skgarden import MondrianForestClassifier
mfc = MondrianForestClassifier()
for i in range(10):
    subset = slice(round(4596729*i),round(4596729*(i+1)))
    mfc.partial_fit(X_set[subset], y_set[subset], classes=np.unique(y_data))

while running the above cell, I get the following error,

AttributeError: ‘MondrianForestClassifier’ object has no attribute ‘partial_fit’

Also there was no method ‘partial_fit’ found while running dir(mfc) in a cell.

So i tried installing scikit-garden in my laptop (locally).

Installing through ‘pip’ or ‘conda’, i get the same error.

ModuleNotFoundError: No module named ‘sklearn.ensemble.forest’

In google colab, I think ‘fit’ is working; but i need ‘partial_fit’ to run out-of-core training process.

is there a way to fix this in google colab?

Source: Python Questions

LEAVE A COMMENT