I would like to add a waterfall chart for random forest model interpretation as is suggested in Chapter 9 of Howard & Gugger, however, I am encountering an error.
This is the code I am running in Colab:
!pip install -Uqq waterfallcharts
waterfall(valid_xs_tindep.columns, contributions[0], threshold = 0.08, rotation_value=45, formatting='{:,.3f}');
And, this is the error I am getting:
NameError: name 'waterfall' is not defined
Help to resolve this issue is appreciated.
-Simon
this is how fastbook imported the lib:
from waterfall_chart import plot as waterfall
Thank you, @miwojc.
DO you know if there is another way I should be installing and importing treeinterpreter other than this,
!pip install -Uqq fastai waterfallcharts treeinterpreter dtreeviz
, to use it in the following way?
predictions, bias, contributions = treeinterpreter.predict(m, row.values)
yeah that should work, just import it after installing
from treeinterpreter import treeinterpreter
Import the waterfall function from waterfallcharts by adding from waterfallcharts import waterfall before using it in your code.