단순선형회귀 방법 1, 방법 2 예제(Sequential api, Function api)

 

 

방법1

import tensorflow as tf
from keras.models import Sequential
from keras.layers import Dense, Activation
from keras import optimizers
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.metrics import r2_score
from keras.optimizers import SGD, RMSprop, Adam

df = pd.read_table('http://www.randomservices.org/random/data/Galton.txt')
df = df.loc[(df['Gender'] == 'M')].drop('Gender', axis=1)
print(df.head(3))

x = df['Father']
y = df['Height']
print(x[:3])
print(y[:3])
print(np.corrcoef(x, y)) # 0.391

# train / test split (7 : 3)
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.3, random_state = 0)
print(x_train.shape, x_test.shape, y_train.shape, y_test.shape) # (325,) (140,) (325,) (140,)

model = Sequential()
model.add(Dense(units=5, input_dim=1, activation='linear'))
model.add(Dense(units=1, activation='linear'))
print(model.summary())

opti = optimizers.Adam(learning_rate=0.01)
model.compile(optimizer=opti, loss='mse', metrics=['mse'])

history = model.fit(x_train, y_train, batch_size=4, epochs=100, verbose=0)
loss_metrics = model.evaluate(x_test, y_test, verbose=0)
print('loss_metrics :', loss_metrics)

print('설명력 :', r2_score(y_test, model.predict(x_test)))
print('실제값 :', y_test[:5].values)
print('예측값 :', model.predict(x_test).flatten()[:5])

# 새로운 값 예측
new_height = [75, 70, 80]
print('새로운 예측값: ', model.predict(new_height).flatten())

# 시각화
import matplotlib.pyplot as plt
plt.rc('font', family='malgun gothic')
plt.plot(x_train, model.predict(x_train), 'b', x_test, y_test, 'ko') # train
plt.show()

plt.plot(x_test, model.predict(x_test), 'b', x_test, y_test, 'ko')  # test
plt.show()

plt.plot(history.history['mse'], label='평균제곱오차')
plt.xlabel('학습 횟수')
plt.show()


<console>
  Family  Father  Mother  Height  Kids
0      1    78.5    67.0    73.2     4
4      2    75.5    66.5    73.5     4
5      2    75.5    66.5    72.5     4
0    78.5
4    75.5
5    75.5
Name: Father, dtype: float64
0    73.2
4    73.5
5    72.5
Name: Height, dtype: float64
[[1.         0.39131736]
 [0.39131736 1.        ]]
(325,) (140,) (325,) (140,)
Model: "sequential"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 dense (Dense)               (None, 5)                 10        
                                                                 
 dense_1 (Dense)             (None, 1)                 6         
                                                                 
=================================================================
Total params: 16
Trainable params: 16
Non-trainable params: 0
_________________________________________________________________
None
loss_metrics : [13.743265151977539, 13.743265151977539]

1/5 [=====>........................] - ETA: 0s
5/5 [==============================] - 0s 500us/step
설명력 : -1.0972842836983117
실제값 : [68.  67.  70.  67.  70.2]

1/5 [=====>........................] - ETA: 0s
5/5 [==============================] - 0s 997us/step
예측값 : [67.28584  70.39301  72.14079  73.111786 69.22783 ]

1/1 [==============================] - ETA: 0s
1/1 [==============================] - 0s 53ms/step
새로운 예측값:  [76.99574 72.14079 81.85071]

 1/11 [=>............................] - ETA: 0s
11/11 [==============================] - 0s 708us/step

1/5 [=====>........................] - ETA: 0s
5/5 [==============================] - 0s 748us/step

 

 

방법2

print('\n--Function api 사용------------------------------')
from keras.layers import Input
from keras.models import Model

inputs = Input(shape=(1,))
output1 = Dense(units=5, activation='linear')(inputs)
outputs = Dense(1, activation='linear')(output1)

model2 = Model(inputs, outputs)
print(model2.summary())

opti = tf.keras.optimizers.Adam(learning_rate=0.001)     
model2.compile(optimizer=opti, loss='mse', metrics=['mse']) 
history = model2.fit(x=x_train, y=y_train, epochs=50, batch_size=4, verbose=0)
loss_metrics2 = model2.evaluate(x=x_test, y=y_test, verbose=0)
print('loss metrics: ', loss_metrics2)
print('실제값 : ', y_test.head().values)
print('예측값 :', model2.predict(x_test).flatten()[:5])
print()

new_data = [75, 70, 80]
print('새로운 예측 키 값: ', model2.predict(new_data).flatten())


<console>
--Function api 사용------------------------------
Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 1)]               0         
                                                                 
 dense_2 (Dense)             (None, 5)                 10        
                                                                 
 dense_3 (Dense)             (None, 1)                 6         
                                                                 
=================================================================
Total params: 16
Trainable params: 16
Non-trainable params: 0
_________________________________________________________________
None
loss metrics:  [8.111572265625, 8.111572265625]
실제값 :  [68.  67.  70.  67.  70.2]

1/5 [=====>........................] - ETA: 0s
5/5 [==============================] - 0s 748us/step
예측값 : [65.24897  68.451355 70.2527   71.25345  67.25046 ]


1/1 [==============================] - ETA: 0s
1/1 [==============================] - 0s 50ms/step
새로운 예측 키 값:  [75.25643 70.2527  80.26017]

 

 

train 시각화

 

 

test 시각화