@@ -138,7 +138,7 @@ cdef class ClassificationCriterion(Criterion):
138138 cdef SIZE_t k = 0
139139 cdef SIZE_t label_count_stride = 0
140140
141- for k from 0 <= k < n_outputs:
141+ for k in range ( n_outputs) :
142142 self .n_classes[k] = n_classes[k]
143143
144144 if n_classes[k] > label_count_stride:
@@ -211,17 +211,17 @@ cdef class ClassificationCriterion(Criterion):
211211 cdef DOUBLE_t w = 1.0
212212 cdef SIZE_t offset = 0
213213
214- for k from 0 <= k < n_outputs:
214+ for k in range ( n_outputs) :
215215 memset(label_count_total + offset, 0 , n_classes[k] * sizeof(double ))
216216 offset += label_count_stride
217217
218- for p from start <= p < end:
218+ for p in range ( start, end) :
219219 i = samples[p]
220220
221221 if sample_weight != NULL :
222222 w = sample_weight[i]
223223
224- for k from 0 <= k < n_outputs:
224+ for k in range ( n_outputs) :
225225 c = < SIZE_t> y[i * y_stride + k]
226226 label_count_total[k * label_count_stride + c] += w
227227
@@ -248,7 +248,7 @@ cdef class ClassificationCriterion(Criterion):
248248
249249 cdef SIZE_t k = 0
250250
251- for k from 0 <= k < n_outputs:
251+ for k in range ( n_outputs) :
252252 memset(label_count_left, 0 , n_classes[k] * sizeof(double ))
253253 memcpy(label_count_right, label_count_total, n_classes[k] * sizeof(double ))
254254
@@ -284,13 +284,13 @@ cdef class ClassificationCriterion(Criterion):
284284
285285 # Note: We assume start <= pos < new_pos <= end
286286
287- for p from pos <= p < new_pos:
287+ for p in range ( pos, new_pos) :
288288 i = samples[p]
289289
290290 if sample_weight != NULL :
291291 w = sample_weight[i]
292292
293- for k from 0 <= k < n_outputs:
293+ for k in range ( n_outputs) :
294294 label_index = (k * label_count_stride +
295295 < SIZE_t> y[i * y_stride + k])
296296 label_count_left[label_index] += w
@@ -318,7 +318,7 @@ cdef class ClassificationCriterion(Criterion):
318318 cdef double * label_count_total = self .label_count_total
319319 cdef SIZE_t k
320320
321- for k from 0 <= k < n_outputs:
321+ for k in range ( n_outputs) :
322322 memcpy(dest, label_count_total, n_classes[k] * sizeof(double ))
323323 dest += label_count_stride
324324 label_count_total += label_count_stride
@@ -354,10 +354,10 @@ cdef class Entropy(ClassificationCriterion):
354354 cdef SIZE_t k
355355 cdef SIZE_t c
356356
357- for k from 0 <= k < n_outputs:
357+ for k in range ( n_outputs) :
358358 entropy = 0.0
359359
360- for c from 0 <= c < n_classes[k]:
360+ for c in range ( n_classes[k]) :
361361 tmp = label_count_total[c]
362362 if tmp > 0.0 :
363363 tmp /= weighted_n_node_samples
@@ -390,11 +390,11 @@ cdef class Entropy(ClassificationCriterion):
390390 cdef SIZE_t k
391391 cdef SIZE_t c
392392
393- for k from 0 <= k < n_outputs:
393+ for k in range ( n_outputs) :
394394 entropy_left = 0.0
395395 entropy_right = 0.0
396396
397- for c from 0 <= c < n_classes[k]:
397+ for c in range ( n_classes[k]) :
398398 tmp = label_count_left[c]
399399 if tmp > 0.0 :
400400 tmp /= weighted_n_left
@@ -445,11 +445,11 @@ cdef class Gini(ClassificationCriterion):
445445 cdef SIZE_t k
446446 cdef SIZE_t c
447447
448- for k from 0 <= k < n_outputs:
448+ for k in range ( n_outputs) :
449449 gini = 0.0
450450
451- for c from 0 <= c < n_classes[k]:
452- tmp = label_count_total[c] # TODO: use weighted count instead
451+ for c in range ( n_classes[k]) :
452+ tmp = label_count_total[c] # TODO: use weighted count instead
453453 gini += tmp * tmp
454454
455455 gini = 1.0 - gini / (weighted_n_node_samples *
@@ -483,12 +483,12 @@ cdef class Gini(ClassificationCriterion):
483483 cdef SIZE_t k
484484 cdef SIZE_t c
485485
486- for k from 0 <= k < n_outputs:
486+ for k in range ( n_outputs) :
487487 gini_left = 0.0
488488 gini_right = 0.0
489489
490- for c from 0 <= c < n_classes[k]:
491- tmp = label_count_left[c] # TODO: use weighted count instead
490+ for c in range ( n_classes[k]) :
491+ tmp = label_count_left[c] # TODO: use weighted count instead
492492 gini_left += tmp * tmp
493493 tmp = label_count_right[c]
494494 gini_right += tmp * tmp
@@ -638,7 +638,7 @@ cdef class RegressionCriterion(Criterion):
638638 cdef DOUBLE_t y_ik = 0.0
639639 cdef DOUBLE_t w = 1.0
640640
641- for k from 0 <= k < n_outputs:
641+ for k in range ( n_outputs) :
642642 mean_left[k] = 0.0
643643 mean_right[k] = 0.0
644644 mean_total[k] = 0.0
@@ -651,13 +651,13 @@ cdef class RegressionCriterion(Criterion):
651651 self .sum_right[k] = 0.0
652652 self .sum_total[k] = 0.0
653653
654- for p from start <= p < end:
654+ for p in range ( start, end) :
655655 i = samples[p]
656656
657657 if sample_weight != NULL :
658658 w = sample_weight[i]
659659
660- for k from 0 <= k < n_outputs:
660+ for k in range ( n_outputs) :
661661 y_ik = y[i * y_stride + k]
662662 sq_sum_total[k] += w * y_ik * y_ik
663663 mean_total[k] += w * y_ik
@@ -667,7 +667,7 @@ cdef class RegressionCriterion(Criterion):
667667
668668 self .weighted_n_node_samples = weighted_n_node_samples
669669
670- for k from 0 <= k < n_outputs:
670+ for k in range ( n_outputs) :
671671 mean_total[k] /= weighted_n_node_samples
672672
673673 # Reset to pos=start
@@ -696,7 +696,7 @@ cdef class RegressionCriterion(Criterion):
696696
697697 cdef SIZE_t k = 0
698698
699- for k from 0 <= k < n_outputs:
699+ for k in range ( n_outputs) :
700700 mean_right[k] = mean_total[k]
701701 mean_left[k] = 0.0
702702 sq_sum_right[k] = sq_sum_total[k]
@@ -738,13 +738,13 @@ cdef class RegressionCriterion(Criterion):
738738
739739 # Note: We assume start <= pos < new_pos <= end
740740
741- for p from pos <= p < new_pos:
741+ for p in range ( pos, new_pos) :
742742 i = samples[p]
743743
744744 if sample_weight != NULL :
745745 w = sample_weight[i]
746746
747- for k from 0 <= k < n_outputs:
747+ for k in range ( n_outputs) :
748748 y_ik = y[i * y_stride + k]
749749 w_y_ik = w * y_ik
750750
@@ -762,7 +762,7 @@ cdef class RegressionCriterion(Criterion):
762762 weighted_n_left += w
763763 weighted_n_right -= w
764764
765- for k from 0 <= k < n_outputs:
765+ for k in range ( n_outputs) :
766766 var_left[k] = (sq_sum_left[k] / weighted_n_left -
767767 mean_left[k] * mean_left[k])
768768 var_right[k] = (sq_sum_right[k] / weighted_n_right -
@@ -799,7 +799,7 @@ cdef class MSE(RegressionCriterion):
799799 cdef double total = 0.0
800800 cdef SIZE_t k
801801
802- for k from 0 <= k < n_outputs:
802+ for k in range ( n_outputs) :
803803 total += (sq_sum_total[k] / weighted_n_node_samples -
804804 mean_total[k] * mean_total[k])
805805
@@ -816,7 +816,7 @@ cdef class MSE(RegressionCriterion):
816816 cdef double total_right = 0.0
817817 cdef SIZE_t k
818818
819- for k from 0 <= k < n_outputs:
819+ for k in range ( n_outputs) :
820820 total_left += var_left[k]
821821 total_right += var_right[k]
822822
@@ -913,7 +913,7 @@ cdef class Splitter:
913913 cdef SIZE_t i, j
914914 j = 0
915915
916- for i from 0 <= i < n_samples:
916+ for i in range ( n_samples) :
917917 # Only work with positively weighted samples
918918 if sample_weight == NULL or sample_weight[i] != 0.0 :
919919 samples[j] = i
@@ -925,7 +925,7 @@ cdef class Splitter:
925925 cdef SIZE_t n_features = X.shape[1 ]
926926 cdef SIZE_t* features = < SIZE_t* > malloc(n_features * sizeof(SIZE_t))
927927
928- for i from 0 <= i < n_features:
928+ for i in range ( n_features) :
929929 features[i] = i
930930
931931 self .features = features
@@ -994,8 +994,8 @@ cdef class BestSplitter(Splitter):
994994 cdef double best_impurity_left = INFINITY
995995 cdef double best_impurity_right = INFINITY
996996 cdef SIZE_t best_pos = end
997- cdef SIZE_t best_feature
998- cdef double best_threshold
997+ cdef SIZE_t best_feature = 0
998+ cdef double best_threshold = 0.
999999 cdef double best_improvement = - INFINITY
10001000
10011001 cdef double current_improvement
@@ -1012,7 +1012,7 @@ cdef class BestSplitter(Splitter):
10121012 cdef SIZE_t partition_start
10131013 cdef SIZE_t partition_end
10141014
1015- for f_idx from 0 <= f_idx < n_features:
1015+ for f_idx in range ( n_features) :
10161016 # Draw a feature at random
10171017 f_i = n_features - f_idx - 1
10181018 f_j = rand_int(n_features - f_idx, random_state)
@@ -1176,8 +1176,8 @@ cdef class RandomSplitter(Splitter):
11761176 cdef double best_impurity_left = INFINITY
11771177 cdef double best_impurity_right = INFINITY
11781178 cdef SIZE_t best_pos = end
1179- cdef SIZE_t best_feature
1180- cdef double best_threshold
1179+ cdef SIZE_t best_feature = 0
1180+ cdef double best_threshold = 0.
11811181 cdef double best_improvement = - INFINITY
11821182
11831183 cdef double current_improvement
@@ -1197,7 +1197,7 @@ cdef class RandomSplitter(Splitter):
11971197 cdef SIZE_t partition_start
11981198 cdef SIZE_t partition_end
11991199
1200- for f_idx from 0 <= f_idx < n_features:
1200+ for f_idx in range ( n_features) :
12011201 # Draw a feature at random
12021202 f_i = n_features - f_idx - 1
12031203 f_j = rand_int(n_features - f_idx, random_state)
@@ -1211,7 +1211,7 @@ cdef class RandomSplitter(Splitter):
12111211 # Find min, max
12121212 min_feature_value = max_feature_value = X[X_sample_stride * samples[start] + X_fx_stride * current_feature]
12131213
1214- for p from start < p < end:
1214+ for p in range ( start + 1 , end) :
12151215 current_feature_value = X[X_sample_stride * samples[p] + X_fx_stride * current_feature]
12161216
12171217 if current_feature_value < min_feature_value:
@@ -1379,8 +1379,8 @@ cdef class PresortBestSplitter(Splitter):
13791379 cdef double best_impurity_left = INFINITY
13801380 cdef double best_impurity_right = INFINITY
13811381 cdef SIZE_t best_pos = end
1382- cdef SIZE_t best_feature
1383- cdef double best_threshold
1382+ cdef SIZE_t best_feature = 0
1383+ cdef double best_threshold = 0.
13841384 cdef double best_improvement = - INFINITY
13851385
13861386 cdef double current_improvement
@@ -1400,11 +1400,11 @@ cdef class PresortBestSplitter(Splitter):
14001400 cdef SIZE_t i, j
14011401
14021402 # Set sample mask
1403- for p from start <= p < end:
1403+ for p in range ( start, end) :
14041404 sample_mask[samples[p]] = 1
14051405
14061406 # Look for splits
1407- for f_idx from 0 <= f_idx < n_features:
1407+ for f_idx in range ( n_features) :
14081408 # Draw a feature at random
14091409 f_i = n_features - f_idx - 1
14101410 f_j = rand_int(n_features - f_idx, random_state)
@@ -1420,7 +1420,7 @@ cdef class PresortBestSplitter(Splitter):
14201420 # Extract ordering from X_argsorted
14211421 p = start
14221422
1423- for i from 0 <= i < n_total_samples:
1423+ for i in range ( n_total_samples) :
14241424 j = X_argsorted[X_argsorted_stride * current_feature + i]
14251425 if sample_mask[j] == 1 :
14261426 samples[p] = j
@@ -1496,7 +1496,7 @@ cdef class PresortBestSplitter(Splitter):
14961496 samples[p] = tmp
14971497
14981498 # Reset sample mask
1499- for p from start <= p < end:
1499+ for p in range ( start, end) :
15001500 sample_mask[samples[p]] = 0
15011501
15021502 # Return values
@@ -1933,7 +1933,7 @@ cdef class Tree:
19331933
19341934 cdef SIZE_t k
19351935
1936- for k from 0 <= k < n_outputs:
1936+ for k in range ( n_outputs) :
19371937 self .n_classes[k] = n_classes[k]
19381938
19391939 # Parameters
@@ -2152,7 +2152,7 @@ cdef class Tree:
21522152 out = np.zeros((n_samples, max_n_classes), dtype = np.float64)
21532153
21542154 with nogil:
2155- for i from 0 <= i < n_samples:
2155+ for i in range ( n_samples) :
21562156 node_id = 0
21572157
21582158 # While node_id not a leaf
@@ -2165,7 +2165,7 @@ cdef class Tree:
21652165
21662166 offset = node_id * value_stride
21672167
2168- for c from 0 <= c < n_classes[0 ]:
2168+ for c in range ( n_classes[0 ]) :
21692169 out[i, c] = value[offset + c]
21702170
21712171 return out
@@ -2176,7 +2176,7 @@ cdef class Tree:
21762176 max_n_classes), dtype = np.float64)
21772177
21782178 with nogil:
2179- for i from 0 <= i < n_samples:
2179+ for i in range ( n_samples) :
21802180 node_id = 0
21812181
21822182 # While node_id not a leaf
@@ -2189,8 +2189,8 @@ cdef class Tree:
21892189
21902190 offset = node_id * value_stride
21912191
2192- for k from 0 <= k < n_outputs:
2193- for c from 0 <= c < n_classes[k]:
2192+ for k in range ( n_outputs) :
2193+ for c in range ( n_classes[k]) :
21942194 out_multi[i, k, c] = value[offset + c]
21952195 offset += max_n_classes
21962196
@@ -2211,7 +2211,7 @@ cdef class Tree:
22112211 out = np.zeros((n_samples,), dtype = np.int32)
22122212
22132213 with nogil:
2214- for i from 0 <= i < n_samples:
2214+ for i in range ( n_samples) :
22152215 node_id = 0
22162216
22172217 # While node_id not a leaf
@@ -2244,7 +2244,7 @@ cdef class Tree:
22442244 cdef np.ndarray[np.float64_t, ndim= 1 ] importances
22452245 importances = np.zeros((self .n_features,))
22462246
2247- for node from 0 <= node < node_count:
2247+ for node in range ( node_count) :
22482248 if children_left[node] != _TREE_LEAF:
22492249 # ... and children_right[node] != _TREE_LEAF:
22502250 n_left = n_node_samples[children_left[node]]
@@ -2279,7 +2279,7 @@ cdef inline UINT32_t our_rand_r(UINT32_t* seed) nogil:
22792279 seed[0 ] ^= < UINT32_t> (seed[0 ] >> 17 )
22802280 seed[0 ] ^= < UINT32_t> (seed[0 ] << 5 )
22812281
2282- return seed[0 ] % < UINT32_t> ( RAND_R_MAX + 1 )
2282+ return seed[0 ] % ( < UINT32_t> RAND_R_MAX + 1 )
22832283
22842284cdef inline np.ndarray int_ptr_to_ndarray(int * data, SIZE_t size):
22852285 """ Encapsulate data into a 1D numpy array of int's."""
0 commit comments