-
Notifications
You must be signed in to change notification settings - Fork 0
/
RPTree.py
170 lines (143 loc) · 7.78 KB
/
RPTree.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
# -*- coding: utf-8 -*-
"""
Created on Mon Sep 20 18:46:56 2021
@author: Mashaan
"""
import numpy as np
import matplotlib.pyplot as plt
from sklearn import random_projection
from sklearn.decomposition import PCA
import random
class Node(object):
def __init__(self, data):
self.data = data
self.hyperplane = None
self.PCAmean = None
self.splitDimension = None
self.splitPoint = None
self.left = None
self.right = None
class BinaryTree(object):
def __init__(self, root):
self.root = Node(root)
def construct_tree(self, tree, whichProjection):
nTry = 3
#whichProjection = '2008_Dasgupta' # '2008_Dasgupta' # '2019_Yan' # 'proposed' # 'PCA'
X_data = tree.root.data
if whichProjection == '2008_Dasgupta':
transformer = random_projection.GaussianRandomProjection(n_components=X_data.shape[1]-1)
X_proj = transformer.fit_transform(X_data)
hyperplane = transformer.components_
elif whichProjection == '2019_Yan':
dispersion = 0
for r in range(nTry):
transformer = random_projection.GaussianRandomProjection(n_components=X_data.shape[1]-1)
X_proj_temp = transformer.fit_transform(X_data)
dispersionCurrent = np.max(np.std(X_proj_temp, axis=0))
# print('dispersionCurrent = ' + str(dispersionCurrent))
if dispersionCurrent > dispersion:
dispersion = dispersionCurrent
# print('dispersion = ' + str(dispersion))
X_proj = X_proj_temp
hyperplane = transformer.components_
# print('=============================================================')
elif whichProjection == 'proposed':
dispersion = 0
for r in range(nTry):
transformer = random_projection.GaussianRandomProjection(n_components=X_data.shape[1]-1)
X_proj_temp = transformer.fit_transform(X_data)
dispersionCurrent = np.max(np.std(X_proj_temp, axis=0))
# print('dispersionCurrent = ' + str(dispersionCurrent))
if dispersionCurrent > dispersion:
dispersion = dispersionCurrent
# print('dispersion = ' + str(dispersion))
X_proj = X_proj_temp
hyperplane = transformer.components_
for r in range(nTry):
hyperplane_temp = hyperplane + np.random.normal(0, 0.1, hyperplane.shape)
X_proj_temp = np.dot(X_data,np.transpose(hyperplane_temp))
dispersionCurrent = np.max(np.std(X_proj_temp, axis=0))
# print('dispersionCurrent = ' + str(dispersionCurrent))
if dispersionCurrent > dispersion:
dispersion = dispersionCurrent
# print('dispersion = ' + str(dispersion))
X_proj = X_proj_temp
hyperplane = hyperplane_temp
for r in range(nTry):
hyperplane_temp = hyperplane + np.random.normal(0, 0.01, hyperplane.shape)
X_proj_temp = np.dot(X_data,np.transpose(hyperplane_temp))
dispersionCurrent = np.max(np.std(X_proj_temp, axis=0))
# print('dispersionCurrent = ' + str(dispersionCurrent))
if dispersionCurrent > dispersion:
dispersion = dispersionCurrent
# print('dispersion = ' + str(dispersion))
X_proj = X_proj_temp
hyperplane = hyperplane_temp
# print('=============================================================')
elif whichProjection == 'PCA':
# this line could throw an error if the number of samples is less than the number of principle components selected
pca = PCA(n_components=min(X_data.shape[0]-1, X_data.shape[1]-1))
pca.fit(X_data)
X_proj = pca.transform(X_data)
hyperplane = pca.components_
tree.root.PCAmean = pca.mean_
SplitDimension = np.argmax(np.std(X_proj, axis=0))
SplitPoint = random.uniform(np.quantile(X_proj[:,SplitDimension], 0.25, axis=0),np.quantile(X_proj[:,SplitDimension], 0.75, axis=0))
X_left = X_data[np.where(X_proj[:,SplitDimension] < SplitPoint)[0]]
X_right = X_data[np.where(X_proj[:,SplitDimension] >= SplitPoint)[0]]
tree.root.hyperplane = hyperplane
tree.root.splitDimension = SplitDimension
tree.root.splitPoint = SplitPoint
# fig=plt.figure(figsize=(6,6))
# ax=fig.add_subplot(111)
# ax.scatter(X_left[:,0], X_left[:,1], c='b')
# ax.scatter(X_right[:,0], X_right[:,1], c='r')
# plt.tick_params(axis='both',which='both',bottom=False,top=False,left=False,right=False,
# labelbottom=False,labeltop=False,labelleft=False,labelright=False)
if X_left.shape[0]>20:
tree.root.left = self.construct_tree(BinaryTree(X_left), whichProjection)
else:
tree.root.left = Node(X_left)
if X_right.shape[0]>20:
tree.root.right = self.construct_tree(BinaryTree(X_right), whichProjection)
else:
tree.root.right = Node(X_right)
return tree.root
def get_leaf_nodes(self):
leafs = []
self._collect_leaf_nodes(self.root,leafs)
return leafs
def _collect_leaf_nodes(self, node, leafs):
if node is not None:
if node.left==None and node.right==None:
leafs.append(node)
self._collect_leaf_nodes(node.left, leafs)
self._collect_leaf_nodes(node.right, leafs)
def preorder_search(self, NodeRoot, NodeSearch):
if NodeRoot.left==None or NodeRoot.right==None:
return NodeRoot.data
else:
# fig=plt.figure(figsize=(6,6))
# ax=fig.add_subplot(111)
# ax.scatter(NodeRoot.left.data[:,0], NodeRoot.left.data[:,1], c='b')
# ax.scatter(NodeRoot.right.data[:,0], NodeRoot.right.data[:,1], c='r')
# ax.scatter(NodeSearch.data[0], NodeSearch.data[1], marker='x', c='k')
# plt.tick_params(axis='both',which='both',bottom=False,top=False,left=False,right=False,
# labelbottom=False,labeltop=False,labelleft=False,labelright=False)
if NodeRoot.PCAmean is not None:
ProjectedPoint = np.dot((NodeSearch.data - NodeRoot.PCAmean),np.transpose(NodeRoot.hyperplane))
else:
ProjectedPoint = np.dot(NodeSearch.data,np.transpose(NodeRoot.hyperplane))
# print('NodeRoot.data.shape = ' + str(NodeRoot.data.shape))
# print('ProjectedPoint = ' + str(ProjectedPoint))
# print('NodeRoot.splitDimension = ' + str(NodeRoot.splitDimension))
# print('ProjectedPoint[NodeRoot.splitDimension] = ' + str(ProjectedPoint[NodeRoot.splitDimension]))
# print('NodeRoot.splitPoint = ' + str(NodeRoot.splitPoint))
if ProjectedPoint[NodeRoot.splitDimension] < NodeRoot.splitPoint:
# print('Im going left')
# print('=============================================================')
return self.preorder_search(NodeRoot.left, NodeSearch)
else:
# print('Im going right')
# print('=============================================================')
return self.preorder_search(NodeRoot.right, NodeSearch)