Reproducing KNN warnings

gistfile1.py

from pprint import pprint

from sklearn.datasets import load_digits
from sklearn.neighbors import KNeighborsClassifier
from sklearn.cross_validation import KFold

# load data
digits = load_digits()
X = digits.data.astype('uint8')
Y = digits.target

# try to predict
scores = []
for (train, test) in KFold(n=X.shape[0], k=5):
  X_train, X_test, Y_train, Y_test = X[train], X[test], Y[train], Y[test]
  clf = KNeighborsClassifier(n_neighbors=1, algorithm='brute')
  
  print 'as type {}'.format(X_train.dtype)
  score1 = clf.fit(X_train, Y_train).score(X_test, Y_test)
  
  print 'as type {}'.format(X_train.astype(float).dtype)
  score2 = clf.fit(X_train.astype(float), Y_train).score(X_test.astype(float), Y_test)
  scores.append( (score1, score2) )
print scores