Update to work with new versions of scipy and scikit

packtml/base.py

@@ -3,8 +3,7 @@
 from __future__ import absolute_import
 
 from abc import ABCMeta, abstractmethod
-
-from sklearn.externals import six
+import six
 
 __all__ = [
     'BaseSimpleEstimator'

packtml/clustering/knn.py

@@ -103,4 +103,4 @@ class KNNClassifier(BaseSimpleEstimator):
         # We want the most common along the rows as the predictions
         # I.e:
         # array([1, ..., 0])
-        return mode(predicted_labels, axis=-1)[0].ravel()
+        return mode(predicted_labels, axis=1)[0].ravel()

packtml/neural_net/base.py

@@ -2,7 +2,7 @@
 
 from __future__ import absolute_import
 
-from sklearn.externals import six
+import six
 from abc import ABCMeta, abstractmethod
 
 import numpy as np

packtml/recommendation/base.py

@@ -2,7 +2,7 @@
 
 from __future__ import absolute_import
 
-from sklearn.externals import six
+import six
 from abc import ABCMeta, abstractmethod
 
 __all__ = [

packtml/regression/simple_regression.py

@@ -70,7 +70,7 @@ class SimpleLinearRegression(BaseSimpleEstimator):
         # Let's compute the least squares on X wrt y
         # Least squares solves the equation `a x = b` by computing a
         # vector `x` that minimizes the Euclidean 2-norm `|| b - a x ||^2`.
-        theta, _, rank, singular_values = lstsq(X, y)
+        theta, _, rank, singular_values = lstsq(X, y, rcond=None)
 
         # finally, we compute the intercept values as the mean of the target
         # variable MINUS the inner product of the X_means and the coefficients

packtml/utils/tests/test_linalg.py

@@ -5,7 +5,6 @@ from __future__ import absolute_import
 from sklearn.datasets import load_iris
 from packtml.utils import linalg
 
-from numpy.testing import assert_array_almost_equal
 import numpy as np
 
 iris = load_iris()
@@ -17,7 +16,7 @@ def test_row_norms():
     X_centered = X - means
 
     norms = linalg.l2_norm(X_centered, axis=0)
-    assert_array_almost_equal(
+    assert np.allclose(
         norms,
-        np.array([ 10.10783524,   5.29269308,
-                   21.53749599,   9.31556404]))
+        np.array([10.10783524, 5.29269308, 21.53749599, 9.31556404]),
+        rtol=0.01)

packtml/utils/validation.py

@@ -3,7 +3,6 @@
 from __future__ import absolute_import
 
 from sklearn.externals import six
-from sklearn.utils.validation import check_random_state
 from sklearn.model_selection import ShuffleSplit
 
 import numpy as np

requirements.txt

@@ -1,5 +1,5 @@
-numpy>=0.11
+numpy>=0.15
 scipy>=0.19
-scikit-learn>=0.18
-pandas
+scikit-learn>=0.19
+pandas>=0.23
 matplotlib