-
Notifications
You must be signed in to change notification settings - Fork 1
/
classification-exp.py
141 lines (119 loc) · 5.28 KB
/
classification-exp.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
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split, KFold
# from HAR.MakeDataset import X_train, y_train
from tree.utils import *
from tree.base import *
from metrics import *
from sklearn.datasets import make_classification
np.random.seed(42)
# Code given in the question
X, y = make_classification(
n_features=2, n_redundant=0, n_informative=2, random_state=1, n_clusters_per_class=2, class_sep=0.5)
# For plotting
plt.scatter(X[:, 0], X[:, 1], c=y)
xtrain, xtest, ytrain, ytest = train_test_split(X[:, 0], y, test_size=0.3)
xtrain = pd.DataFrame(xtrain)
ytrain = pd.Series(ytrain)
ytest = pd.Series(ytest)
xtest = pd.DataFrame(xtest)
tree = DecisionTree(criterion="information_gain")
tree.fit(xtrain, ytrain)
y_hat = tree.predict(xtest)
print("Accuracy of our Decision tree on first class is " + str(accuracy(ytest, y_hat) * 100) + "%")
for cls in ytest.unique():
print(f"Precision:{precision(y_hat, ytest, cls)}")
print(f"Recall: {recall(y_hat, ytest, cls)}")
xtrain2, xtest2, ytrain2, ytest2 = train_test_split(X[:, 1], y, test_size=0.3)
xtrain2 = pd.DataFrame(xtrain2)
ytrain2 = pd.Series(ytrain2)
ytest2 = pd.Series(ytest2)
xtest2 = pd.DataFrame(xtest2)
tree = DecisionTree(criterion="information_gain")
tree.fit(xtrain2, ytrain2)
y_hat = tree.predict(xtest2)
print("Accuracy of our Decision tree on second class is " + str(accuracy(y_hat, ytest2) * 100) + "%")
for cls in ytest2.unique():
print(f"Precision:{precision(y_hat, ytest2, cls)}")
print(f"Recall: {recall(y_hat, ytest2, cls)}")
n_folds = 5
dataset_size = len(X)
indices = np.arange(dataset_size)
np.random.shuffle(indices)
fold_sizes = np.full(n_folds, dataset_size // n_folds)
fold_sizes[:dataset_size % n_folds] += 1
current = 0
folds = []
for fold_size in fold_sizes:
start, stop = current, current + fold_size
folds.append((indices[start:stop]))
current = stop
fold_accuracies = []
print("For first class:")
for i in range(n_folds):
# Create the training and validation sets
train_indices = np.concatenate([folds[j] for j in range(n_folds) if j != i])
val_indices = folds[i]
X_train, y_train = pd.DataFrame(X[train_indices, 0]), pd.Series(y[train_indices])
X_val, y_val = pd.DataFrame(X[val_indices, 0]), pd.Series(y[val_indices])
tree.fit(X_train, y_train)
acc = sum(y_val == tree.predict(X_val)) / len(y_val)
fold_accuracies.append(acc)
print(f"Fold {i + 1}, Accuracy: {acc:.4f}")
print(f"Mean accuracy is {np.mean(fold_accuracies)}")
print("For second class:")
fold_accuracies = []
for i in range(n_folds):
# Create the training and validation sets
train_indices = np.concatenate([folds[j] for j in range(n_folds) if j != i])
val_indices = folds[i]
X_train, y_train = pd.DataFrame(X[train_indices, 1]), pd.Series(y[train_indices])
X_val, y_val = pd.DataFrame(X[val_indices, 1]), pd.Series(y[val_indices])
tree.fit(X_train, y_train)
acc = sum(y_val == tree.predict(X_val)) / len(y_val)
fold_accuracies.append(acc)
print(f"Fold {i + 1}, Accuracy: {acc:.4f}")
print(f"Mean accuracy is {np.mean(fold_accuracies)}")
def splitfold(x, n):
i = np.arange(len(x))
np.random.shuffle(i)
f = np.array_split(i, n)
return f
def nestedvalid(x, y,cl, depths=[2, 3, 4, 5, 6], ofold=5, ifold=5):
oscore = []
ofold_indices = splitfold(x, ofold)
for outer_train_indices in ofold_indices:
outer_test_indices = np.concatenate(
[outer_train_indices for idx in range(ofold) if not np.array_equal(outer_train_indices, idx)])
X_train, X_test = X[outer_train_indices, cl], X[outer_test_indices, cl]
y_train, y_test = y[outer_train_indices], y[outer_test_indices]
inner_folds_indices = splitfold(X_train, ifold)
best_depth = None
best_score = -np.inf
for depth in depths:
iscore = []
for inner_train_indices in inner_folds_indices:
inner_val_indices = np.concatenate(
[inner_train_indices for idx in range(ifold) if not np.array_equal(inner_train_indices, idx)])
X_inner_train, X_inner_val = X_train[inner_train_indices], X_train[inner_val_indices]
y_inner_train, y_inner_val = y_train[inner_train_indices], y_train[inner_val_indices]
tree = DecisionTree(criterion="information_gain", maxDepth=depth)
tree.fit(X_inner_train, y_inner_train)
acc = sum(y_inner_val == tree.predict(X_inner_val)) / len(y_inner_val)
iscore.append(acc)
mean_score = np.mean(iscore)
if mean_score > best_score:
best_score = mean_score
best_depth = depth
final_model = DecisionTree(criterion='information_gain', maxDepth=best_depth)
final_model.fit(X_train, y_train)
y_pred = final_model.predict(X_test)
oscore.append(accuracy(y_test, y_pred))
return np.mean(oscore) * 100, np.std(oscore)
mscore, stdcore = nestedvalid(X, y,0, depths=[2, 3, 4, 5, 6], ofold=5, ifold=5)
print(str(mscore) + " % Accuracy for first class")
print("Standard Deviation is " + str(stdcore))
mscore, stdcore = nestedvalid(X, y,1, depths=[2, 3, 4, 5, 6], ofold=5, ifold=5)
print(str(mscore) + " % Accuracy for second class")
print("Standard Deviation is " + str(stdcore))