TreeEnsembleClassifier - 1 vs 3#

Next section compares an older to a newer version of the same operator after both definition are converted into markdown text. Green means an addition to the newer version, red means a deletion. Anything else is unchanged.

Files changed (1) hide show
  1. TreeEnsembleClassifier1 → TreeEnsembleClassifier3 +1 -13
TreeEnsembleClassifier1 → TreeEnsembleClassifier3 RENAMED
@@ -1 +1 @@
1
- Tree Ensemble classifier. Returns the top class for each of N inputs.
1
+ Tree Ensemble classifier. Returns the top class for each of N inputs.
2
2
  The attributes named 'nodes_X' form a sequence of tuples, associated by
3
3
  index into the sequences, which must all be of equal length. These tuples
4
4
  define the nodes.
5
5
  Similarly, all fields prefixed with 'class_' are tuples of votes at the leaves.
6
6
  A leaf may have multiple votes, where each vote is weighted by
7
7
  the associated class_weights index.
8
8
  One and only one of classlabels_strings or classlabels_int64s
9
9
  will be defined. The class_ids are indices into this list.
10
- All fields ending with <i>_as_tensor</i> can be used instead of the
11
- same parameter without the suffix if the element type is double and not float.
12
10
  **Attributes**
13
11
  * **base_values**:
14
- Base values for classification, added to final class score; the size
15
- must be the same as the classes or can be left unassigned (assumed
16
- 0)
17
- * **base_values_as_tensor**:
18
12
  Base values for classification, added to final class score; the size
19
13
  must be the same as the classes or can be left unassigned (assumed
20
14
  0)
21
15
  * **class_ids**:
22
16
  The index of the class list that each weight is for.
23
17
  * **class_nodeids**:
24
18
  node id that this weight is for.
25
19
  * **class_treeids**:
26
20
  The id of the tree that this node is in.
27
21
  * **class_weights**:
28
- The weight for the class in class_id.
29
- * **class_weights_as_tensor**:
30
22
  The weight for the class in class_id.
31
23
  * **classlabels_int64s**:
32
24
  Class labels if using integer labels.<br>One and only one of the
33
25
  'classlabels_*' attributes must be defined.
34
26
  * **classlabels_strings**:
35
27
  Class labels if using string labels.<br>One and only one of the
36
28
  'classlabels_*' attributes must be defined.
37
29
  * **nodes_falsenodeids**:
38
30
  Child node if expression is false.
39
31
  * **nodes_featureids**:
40
32
  Feature id for each node.
41
33
  * **nodes_hitrates**:
42
- Popularity of each node, used for performance and may be omitted.
43
- * **nodes_hitrates_as_tensor**:
44
34
  Popularity of each node, used for performance and may be omitted.
45
35
  * **nodes_missing_value_tracks_true**:
46
36
  For each node, define what to do in the presence of a missing value:
47
37
  if a value is missing (NaN), use the 'true' or 'false' branch based
48
38
  on the value in this array.<br>This attribute may be left undefined,
49
39
  and the defalt value is false (0) for all nodes.
50
40
  * **nodes_modes**:
51
41
  The node kind, that is, the comparison to make at the node. There is
52
42
  no comparison to make at a leaf node.<br>One of 'BRANCH_LEQ',
53
43
  'BRANCH_LT', 'BRANCH_GTE', 'BRANCH_GT', 'BRANCH_EQ', 'BRANCH_NEQ',
54
44
  'LEAF'
55
45
  * **nodes_nodeids**:
56
46
  Node id for each node. Ids may restart at zero for each tree, but it
57
47
  not required to.
58
48
  * **nodes_treeids**:
59
49
  Tree id for each node.
60
50
  * **nodes_truenodeids**:
61
51
  Child node if expression is true.
62
52
  * **nodes_values**:
63
- Thresholds to do the splitting on for each node.
64
- * **nodes_values_as_tensor**:
65
53
  Thresholds to do the splitting on for each node.
66
54
  * **post_transform**:
67
55
  Indicates the transform to apply to the score. <br> One of 'NONE,'
68
56
  'SOFTMAX,' 'LOGISTIC,' 'SOFTMAX_ZERO,' or 'PROBIT.'
69
57
  **Inputs**
70
58
  * **X** (heterogeneous) - **T1**:
71
59
  Input of shape [N,F]
72
60
  **Outputs**
73
61
  * **Y** (heterogeneous) - **T2**:
74
62
  N, Top class for each point
75
63
  * **Z** (heterogeneous) - **tensor(float)**:
76
64
  The class score for each class, for each point, a tensor of shape
77
65
  [N,E].
78
66
  **Type Constraints**
79
67
  * **T1** in (
80
68
  tensor(double),
81
69
  tensor(float),
82
70
  tensor(int32),
83
71
  tensor(int64)
84
72
  ):
85
73
  The input type must be a tensor of a numeric type.
86
74
  * **T2** in (
87
75
  tensor(int64),
88
76
  tensor(string)
89
77
  ):
90
78
  The output type will be a tensor of strings or integers, depending
91
79
  on which of the classlabels_* attributes is used.