I am trying to implement a cross validation scheme on grouped data. I was hoping to use the GroupKFold method, but I keep getting an error. what am I doing wrong? The code (slightly different from the one I used--I had different data so I had a larger n_splits, but everythign else is the same)
from sklearn import metrics
import matplotlib.pyplot as plt
import numpy as np
from sklearn.model_selection import GroupKFold
from sklearn.grid_search import GridSearchCV
from xgboost import XGBRegressor
#generate data
x=np.array([0,1,2,3,4,5,6,7,8,9,10,11,12,13])
y= np.array([1,2,3,4,5,6,7,1,2,3,4,5,6,7])
group=np.array([1,0,1,1,2,2,2,1,1,1,2,0,0,2)]
#grid search
gkf = GroupKFold( n_splits=3).split(x,y,group)
subsample = np.arange(0.3,0.5,0.1)
param_grid = dict( subsample=subsample)
rgr_xgb = XGBRegressor(n_estimators=50)
grid_search = GridSearchCV(rgr_xgb, param_grid, cv=gkf, n_jobs=-1)
result = grid_search.fit(x, y)
the error:
Traceback (most recent call last):
File "<ipython-input-143-11d785056a08>", line 8, in <module>
result = grid_search.fit(x, y)
File "/home/student/anaconda/lib/python3.5/site-packages/sklearn/grid_search.py", line 813, in fit
return self._fit(X, y, ParameterGrid(self.param_grid))
File "/home/student/anaconda/lib/python3.5/site-packages/sklearn/grid_search.py", line 566, in _fit
n_folds = len(cv)
TypeError: object of type 'generator' has no len()
changing the line
gkf = GroupKFold( n_splits=3).split(x,y,group)
to
gkf = GroupKFold( n_splits=3)
does not work either. The error message is then:
'GroupKFold' object is not iterable
The split
function of GroupKFold
yields the training and test indices pair one at a time. You should call list
on the split value to get them all in a list so the length can be computed:
gkf = list(GroupKFold( n_splits=3).split(x,y,group))
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With