-
Notifications
You must be signed in to change notification settings - Fork 0
/
DAI_ASSIGNMENT_1_Q3.py
255 lines (203 loc) · 543 KB
/
DAI_ASSIGNMENT_1_Q3.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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
# -*- coding: utf-8 -*-
"""DAI_ASSIGNMENT_1_Q3.ipynb
Automatically generated by Colaboratory.
Original file is located at
https://colab.research.google.com/drive/1p-hIhXw4nULWbEaa9xPKbUspzuFZOdJx
Q3. Use the CIFAR10 dataset, take 5 classes and perform the classification. Print confusion matrix for 5 classes. [Download Dataset from here] [ 20 Marks]
Note: Code for Confusion matrix must be done from scratch, else no marks will be awarded.
"""
from tensorflow.keras import datasets
import numpy as np
filtered_images=[]
(train_images, train_labels), (test_images, test_labels)= datasets.cifar10.load_data()
print('Images Shape: {}'.format(train_images.shape))
print('Labels Shape: {}'.format(train_labels.shape))
for i in range(0,5):
idx = (train_labels == i).reshape(train_images.shape[0])
# print('Index Shape: {}'.format(idx.shape))
filtered_images.append(train_images[idx])
print('Filtered Images Shape: {}'.format(np.array(filtered_images).shape))
# Train Images
Xtrain_airplane = np.array(filtered_images[0])
Xtrain_automobile = np.array(filtered_images[1])
Xtrain_bird = np.array(filtered_images[2])
Xtrain_cat = np.array(filtered_images[3])
Xtrain_deer = np.array(filtered_images[4])
filtered_labels = []
for i in range(0,5):
idx = (train_labels == i).reshape(train_labels.shape[0])
filtered_labels.append(train_labels[idx])
print('Filtered labels Shape: {}'.format(np.array(filtered_labels).shape))
# Train Labels
Ytrain_airplane = filtered_labels[0]
Ytrain_automobile = filtered_labels[1]
Ytrain_bird = filtered_labels[2]
Ytrain_cat = filtered_labels[3]
Ytrain_deer = filtered_labels[4]
filtered_images_test = []
for i in range(0,5):
idx = (test_labels == i).reshape(test_images.shape[0])
# print('Index Shape: {}'.format(idx.shape))
filtered_images_test.append(test_images[idx])
print('Filtered Images Shape: {}'.format(np.array(filtered_images_test).shape))
# Test Images
Xtest_airplane = np.array(filtered_images_test[0])
Xtest_automobile = np.array(filtered_images_test[1])
Xtest_bird = np.array(filtered_images_test[2])
Xtest_cat = np.array(filtered_images_test[3])
Xtest_deer = np.array(filtered_images_test[4])
filtered_labels_test = []
for i in range(0,5):
idx = (test_labels == i).reshape(test_labels.shape[0])
filtered_labels_test.append(test_labels[idx])
print('Filtered labels Shape: {}'.format(np.array(filtered_labels_test).shape))
# Test Labels
Ytest_airplane = filtered_labels_test[0]
Ytest_automobile = filtered_labels_test[1]
Ytest_bird = filtered_labels_test[2]
Ytest_cat = filtered_labels_test[3]
Ytest_deer = filtered_labels_test[4]
X_train = np.concatenate((Xtrain_airplane,
Xtrain_automobile,
Xtrain_bird,
Xtrain_cat,
Xtrain_deer))
Y_train = np.concatenate((Ytrain_airplane,
Ytrain_automobile,
Ytrain_bird,
Ytrain_cat,
Ytrain_deer))
X_test = np.concatenate((Xtest_airplane,
Xtest_automobile,
Xtest_bird,
Xtest_cat,
Xtest_deer))
Y_test = np.concatenate((Ytest_airplane,
Ytest_automobile,
Ytest_bird,
Ytest_cat,
Ytest_deer))
print(X_train.shape)
print(Y_train.shape)
print(X_test.shape)
print(Y_test.shape)
"""Hurray! We have completed the preprocessing of the data"""
classes = ["airplane",#0
"automobile",#1
"bird",#2
"cat",#3
"deer",#4
]
"""from question 1 we have learned that we will be dealing with class labels and plotting images multiple times. Thus we can define a function which will handle label and image showing.
"""
import matplotlib.pyplot as plt
import numpy as np
def show_label_and_image(X,Y,index):
print("The image in training set is desiplayed as:")
plt.imshow(np.array(X[index]))
print("The label associated with this image is in training set:")
print(Y[index][0])
print(classes[Y[index][0]])
show_label_and_image(X_train, Y_train, 4)
X_train = X_train / 255.0
X_test = X_test / 255.0
"""Considering 5 classes as: "airplane",#0
"automobile",#1
"bird",#2
"cat",#3
"deer",#4
"""
import tensorflow as tf
from tensorflow.keras import layers, models
ann = models.Sequential([
layers.Flatten(input_shape=(32,32,3)),
layers.Dense(256, activation='relu'),
layers.Dense(256, activation='relu'),
layers.Dense(10, activation='sigmoid')
])
ann.compile(optimizer='SGD',
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
ann.fit(X_train, Y_train, epochs=10)
"""# Evaluation Metrics using inbuilt function"""
from sklearn.metrics import confusion_matrix , classification_report
import numpy as np
y_pred = ann.predict(X_test)
y_pred_classes = [np.argmax(element) for element in y_pred]
print("Classification Report: \n", classification_report(Y_test, y_pred_classes))
print("Confusion Matrix: \n", confusion_matrix(Y_test, y_pred_classes))
print("along x axis - predicted, along y axis - Actual")
import seaborn as sns
matrix = confusion_matrix(Y_test, y_pred_classes)
sns.heatmap(matrix, square=True, annot=True, fmt='d', cbar=False,
xticklabels=classes,
yticklabels=classes)
plt.xlabel('true label')
plt.ylabel('predicted label');
"""# Evaluation Metrics and Confusion matrix
without using inbuilt function
"""
# This is what we have as of now
Y_actual = Y_test
Y_pred = y_pred_classes
print(len(Y_actual))
print(len(Y_pred))
# np.array(Y_actual[:,0])
Y_pred
import pandas as pd
data = {
'Y_actual': Y_actual[:,0],
'Y_pred': Y_pred
}
df = pd.DataFrame(data, columns=['Y_actual','Y_pred'])
count=0
for i in range(5000):
if(df.Y_actual.iloc[i] == df.Y_pred.iloc[i]):
count=count+1
else:
pass
print(count)
print("out of 5000 are correctly classified")
confusion_matrix = pd.crosstab(df['Y_pred'], df['Y_actual'], rownames=['Predicted'], colnames=['Actual'])
sns.heatmap(confusion_matrix, annot=True, square=True, fmt='d', cbar=False,)
plt.show()
confusion_matrix
Y_actual = Y_actual[:,0].tolist()
Y_pred = Y_pred
rows, cols = (5, 5)
labels = np.array([0,1,2,3,4])
# arr = [[0]*cols]*rows
arr = np.zeros((len(labels), len(labels)))
for i in range(rows):
for j in range(cols):
arr[i,j] = np.sum((Y_actual == labels[i]) & (Y_pred == labels[j]))
np.array(arr).T
# # Dont know why this did not work????
# rows, cols = (5, 5)
# labels = np.array([0,1,2,3,4])
# arr = [[0]*cols]*rows
# # arr = np.zeros((len(labels), len(labels)))
# for i in range(rows):
# for j in range(cols):
# arr[i][j] = np.sum((Y_actual == labels[i]) & (Y_pred == labels[j]))
# np.array(arr).T
# This will waste a lot of time, dont go with this approach :(
# # cifar10_list[0][1][i][0]
# for i in range (0,50000):
# if(cifar10_list[0][1][i][0]>4):
# np.delete(cifar10_list[0][1], i)
# np.delete(cifar10_list[0][0], i)
# print("deleted", cifar10_list[0][1][i][0])
# # cifar10_list[0][1]
# # cifar10_list[0][0]
# print(len(cifar10_list[0][1]), " in Train Dataset after deletion")
# This will waste a lot of time, dont go with this approach :(
# # cifar10_list[1][1][i][0]
# for i in range (0,10000):
# if(cifar10_list[1][1][i][0]>4):
# np.delete(cifar10_list[1][1], i)
# np.delete(cifar10_list[1][0], i)
# # cifar10_list[0][1]
# # cifar10_list[0][0]
# print(len(cifar10_list[1][1]), " in Test Dataset after deletion")