print the estimated tree model size

libmultilabel/linear/tree.py

@@ -7,6 +7,7 @@
 import sklearn.cluster
 import sklearn.preprocessing
 from tqdm import tqdm
+import psutil
 
 from . import linear
 
@@ -135,13 +136,25 @@ def train_tree(
     root = _build_tree(label_representation, np.arange(y.shape[1]), 0, K, dmax)
 
     num_nodes = 0
+    label_feature_used = ((y.T != 0) * (x != 0)).tocsr()
 
     def count(node):
         nonlocal num_nodes
         num_nodes += 1
+        node.num_nnz_feat = np.count_nonzero(label_feature_used[node.label_map,:].sum(axis=0) != 0)
 
     root.dfs(count)
 
+    # Calculate the total memory (excluding swap) on the local machine
+    total_memory = psutil.virtual_memory().total 
+    print(f'{total_memory / (1024**3):.3f} GB')
+
+    model_size = get_estimated_model_size(root, num_nodes)
+    print(f'*** model_size: {model_size / (1024**3):.3f} GB')
+
+    if (total_memory <= model_size):
+        raise MemoryError(f'Not enough memory to train the model. model_size: {model_size / (1024**3):.3f} GB')
+
     pbar = tqdm(total=num_nodes, disable=not verbose)
 
     def visit(node):
@@ -195,6 +208,26 @@ def _build_tree(label_representation: sparse.csr_matrix, label_map: np.ndarray,
     return Node(label_map=label_map, children=children)
 
 
+def get_estimated_model_size(root, num_nodes):
+    num_nnz_feat, num_branches = np.zeros(num_nodes), np.zeros(num_nodes)
+    num_nodes = 0
+    def collect_stat(node: Node):
+        nonlocal num_nodes
+        num_nnz_feat[num_nodes] = node.num_nnz_feat
+
+        if node.isLeaf():
+            num_branches[num_nodes] = len(node.label_map)
+        else:
+            num_branches[num_nodes] = len(node.children)
+
+        num_nodes += 1
+
+    root.dfs(collect_stat)
+
+    # 16 is because when storing sparse matrices, indices (int64) require 8 bytes and floats require 8 bytes
+    return np.dot(num_nnz_feat, num_branches) * 16
+
+
 def _train_node(y: sparse.csr_matrix, x: sparse.csr_matrix, options: str, node: Node):
     """If node is internal, computes the metalabels representing each child and trains
     on the metalabels. Otherwise, train on y.

requirements.txt

@@ -5,3 +5,4 @@ PyYAML
 scikit-learn
 scipy
 tqdm
+psutil