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
| #%%
import pandas as pd
import numpy as np
from pandas import DataFrame
data = pd.read_csv("D:\\KAGGLE\\Titanic\\train.csv")
def MaxMinNormalization(x):
Max = np.max(x)
Min = np.min(x)
result = (x - Min) / (Max - Min)
return result
# 基本数据处理
data['Sex'] = data['Sex'].apply(lambda s:1 if s == 'male' else 0)
data['Age'] = data.Age.fillna(data.Age.mean())
data['Age'] = MaxMinNormalization(data['Age'])
data['Fare'] = MaxMinNormalization(data['Fare'])
data['Dead'] = data['Survived'].apply(lambda s:int(not s))
data = data.join(pd.get_dummies(data.Embarked, prefix='Embarked'))
# data.drop(['Embarked'], axis=1, inplace=True)
data.info()
data_x = data.filter(regex='Pclass|Sex|Fare|Embarked_.*|Age|SibSp|Parch')
# data_x = data[['Pclass', 'Sex', 'Fare', 'Embarked', 'Age', 'SibSp', 'Parch']]
data_y = data[['Dead', 'Survived']]
from sklearn.model_selection import train_test_split
train_X, test_X, train_Y, test_Y = train_test_split(data_x, data_y, test_size=0.2, random_state=1)
train_X.head()
#%%
# 训练模型
import tensorflow as tf
learning_rate = 0.003
training_epochs = 1000
display_step = (int)(training_epochs / 10)
batch_size = 50
data_width = train_X.shape[1]
data_len = train_X.shape[0]
x = tf.placeholder(tf.float32, [None, data_width])
y = tf.placeholder(tf.float32, [None, 2])
W = tf.Variable(tf.zeros([data_width, 2]))
b = tf.Variable(tf.zeros([2]))
# 预测函数
pred = tf.nn.softmax(tf.matmul(x, W) + b) # Softmax
# loss函数
cost = tf.reduce_mean(-tf.reduce_sum(y*tf.log(pred), reduction_indices=1))
optimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(cost)
init = tf.global_variables_initializer()
saver = tf.train.Saver()
# Start training
with tf.Session() as sess:
# Run the initializer
sess.run(init)
# Training cycle
for epoch in range(training_epochs):
avg_cost = 0.
total_batch = int(data_len/batch_size)
# Loop over all batches
for i in range(total_batch):
if i == total_batch-1:
batch_xs = train_X[i*batch_size:data_len]
batch_ys = train_Y[i*batch_size:data_len]
else:
batch_xs = train_X[i*batch_size:(i+1)*batch_size]
batch_ys = train_Y[i*batch_size:(i+1)*batch_size]
# Run optimization op (backprop) and cost op (to get loss value)
_, c = sess.run([optimizer, cost], feed_dict={x: batch_xs,
y: batch_ys})
# Compute average loss 平均cost
avg_cost += c / total_batch
# Display logs per epoch step
if (epoch+1) % display_step == 0:
print("Epoch:", '%04d' % (epoch+1), "cost=", "{:.9f}".format(avg_cost))
print("Optimization Finished!")
# Test model
correct_prediction = tf.equal(tf.argmax(pred, 1), tf.argmax(y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
print("Accuracy:", accuracy.eval({x: test_X, y: test_Y}))
# save model
saver.save(sess, "D:\KAGGLE\Titanic\save\model.ckpt")
#%%
# 预测结果
data = pd.read_csv("D:\\KAGGLE\\Titanic\\test.csv")
data['Sex'] = data['Sex'].apply(lambda s:1 if s == 'male' else 0)
data['Age'] = data.Age.fillna(data.Age.mean())
data['Age'] = MaxMinNormalization(data['Age'])
data['Fare'] = MaxMinNormalization(data['Fare'])
data = data.join(pd.get_dummies(data.Embarked, prefix='Embarked'))
data_x = data.filter(regex='Pclass|Sex|Fare|Embarked_.*|Age|SibSp|Parch')
with tf.Session() as sess:
# Run the initializer
sess.run(init)
# restore model
saver.restore(sess, "D:\KAGGLE\Titanic\save\model.ckpt")
# Test model
prediction = sess.run(pred, feed_dict={x:data_x})
Survived = np.argmax(prediction, 1)
submission = pd.DataFrame({
"PassengerId": data["PassengerId"],
"Survived": Survived
})
submission.to_csv("D:\\KAGGLE\\Titanic\\titanic_submission.csv", index=False)
|