def get_node(self, node_id):
c = self.conn.cursor()
c.execute("SELECT str_id, host_ip, connected,\
last_conn_time, node_data\
FROM nodes\
WHERE str_id=%s", (node_id,))
row = c.fetchone()
if not row:
return None
str_id, host_ip, connected, last_conn_time, node_data = row[0]
node = Node(str_id)
node.ip_addr = host_ip
node.connected = connected
node.last_conn_time = last_conn_time
if node_data:
node.load_data(node_data)
return node
class Tree():
'''Class for a binary decision tree'''
def __init__(self, tree_id, max_depth, impurity, ndims, min_leaf, num_tests,
num_classes, decision_fn=hardSplit):
self.tree_id = tree_id
self.max_depth = max_depth
self.impurity = impurity
self.ndims = ndims
self.min_leaf = min_leaf
self.num_tests = num_tests
self.num_classes = num_classes
self.decision_fn = decision_fn
# Create root node
dct = {'node_id' : 0,
'residual_depth' : self.max_depth,
'parent' : None,
'impurity' : impurity,
'min_leaf' : min_leaf,
'ndims' : ndims,
'num_tests' : num_tests,
'num_classes' : num_classes,
'decision_fn' : decision_fn}
self.root = Node(**dct)
def train(self, X, y):
'''Build a randomTree'''
self.root.load_data(X, y)
done = False
node = self.root
N = node.train()
return N
def apply(self, X):
'''Propagate the data through the tree'''
marker = numpy.arange(X.shape[0])
result = numpy.asarray(self.root.apply(X, marker))
result = result[numpy.argsort(result[:,0]),1]
return result
