Keras Regression

#!/usr/bin/env python

# coding: utf-8

​

# In[1]:

​

​

#0. 사용할 패키지 불러오기

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

from keras.models import Sequential

from keras.layers import Dense

from keras.layers.recurrent import LSTM

from tensorflow import keras

import tensorflow as tf

import seaborn as sns

​

​

# In[2]:

​

​

#1. 데이터 준비하기

df = pd.read_csv('price.csv')

df.isna().sum()

​

​

# In[3]:

​

​

df.pop('date')

df.pop('in')

df.dropna()

df.head(3)

​

mean(°C)	high(°C)	low(°C)	prep(mm)	humidity(%)	whole_price	retail_price
0	1.2	-3.3	4.0	0.0	73.0	26000	4125
1	5.7	1.0	9.5	0.0	76.9	26000	4125
2	6.5	5.1	9.4	0.0	80.6	26000	4125

# In[4]:

​

​

#sns.pairplot(df[["mean(°C)","high(°C)","low(°C)","prep(mm)","humidity(%)","in","whole_price","retail_price"]], diag_kind="kde")

​

​

# In[5]:

​

​

train_dataset = df.sample(frac=0.8,random_state=0)

test_dataset = df.drop(train_dataset.index)

​

train_stats = df.describe()

train_stats.pop("retail_price")

train_stats = train_stats.transpose()

train_stats

count	mean	std	min	25%	50%	75%	max
mean(°C)	1428.0	13.590126	10.888134	-14.8	4.375	15.0	23.000	33.7
high(°C)	1428.0	9.391737	11.028528	-18.0	0.200	10.2	18.925	30.3
low(°C)	1427.0	18.614085	11.000782	-10.7	9.350	20.6	27.850	39.6
prep(mm)	1428.0	3.065616	11.846932	0.0	0.000	0.0	0.200	144.5
humidity(%)	1428.0	57.811975	14.807831	21.8	46.800	57.8	67.650	97.0
whole_price	1428.0	26810.224090	8658.011133	10000.0	20000.000	26000.0	32250.000	55000.0

​

# In[6]:

​

​

train_dataset = df.sample(frac=0.8,random_state=0)

test_dataset = df.drop(train_dataset.index)

​

train_labels = train_dataset.pop("retail_price")

test_labels = test_dataset.pop("retail_price")

​

​

# In[7]:

​

​

def norm(x):

return (x - train_stats['mean']) / train_stats['std']

​

​

# In[8]:

​

​

normed_train_data = norm(train_dataset)

normed_test_data = norm(test_dataset)

​

#2. 데이터셋 생성하기

# y = train_labels.values # train_dataset[:,-1] #8번째 데이터를 y로

# x = normed_train_data # dataset[:,[0,1,2,4,5]] #0번째부터 7번째 데이터 까지 x로

​

x_train = normed_train_data.values # x[0:850,] # 699번까지

y_train = train_labels.values # y[0:850,]

x_test = normed_test_data.values # x[850:,] #700번부터

y_test = test_labels.values # y[850:,]

​

​

# In[9]:

​

​

# 에포크가 끝날 때마다 점(.)을 출력해 훈련 진행 과정을 표시합니다

class PrintDot(keras.callbacks.Callback):

def on_epoch_end(self, epoch, logs):

if epoch % 100 == 0: print('')

print('.', end='')

​

​

# In[10]:

​

​

act_relu = False

if act_relu:

model = keras.Sequential([

keras.layers.Dense(32, activation='relu', input_shape=[len(train_dataset.keys())]),

keras.layers.Dense(32, activation='relu'),

keras.layers.Dense(1)

])

else:

model = keras.Sequential([

keras.layers.Dense(32, activation=tf.nn.relu, input_shape=[len(train_dataset.keys())]),

keras.layers.Dense(32, activation=tf.nn.relu),

keras.layers.Dense(1)

])

​

​

# In[11]:

​

​

#4. 모델 학습과정 설정하기

adam_opt = False

if adam_opt:

model.compile(loss='mse', optimizer='adam', metrics=['mse', 'binary_crossentropy'])

adam = keras.optimizers.Adam(learning_rate=0.01, beta_1=0.9, beta_2=0.999, amsgrad=False)

model.compile(loss='mse', optimizer=adam, metrics=['mse', 'binary_crossentropy'])

else:

optimizer = tf.keras.optimizers.RMSprop(0.001)

model.compile(loss='mse', optimizer=optimizer, metrics=['mae', 'mse'])

​

#5. 모델 학습시키기

#hist = model.fit(x_train, y_train, epochs=50, batch_size=50)

#hist = model.fit(x_train, y_train, epochs=1000, batch_size=1000, validation_split = 0.2, verbose=0, callbacks=[PrintDot()])

history = model.fit( normed_train_data, train_labels, epochs=100, validation_split = 0.2, verbose=0, callbacks=[PrintDot()])

​

#6. 모델 평가하기

scores = model.evaluate(normed_test_data,y_test)

​

​

# In[12]:

​

​

print(scores)

[678038.5297202797, 578.2003, 678038.5]

​

# In[13]:

​

​

fig, loss_ax=plt.subplots()

loss_ax.plot(history.history['loss'], 'y',label='train_loss')

loss_ax.set_xlabel('epoch')

loss_ax.set_ylabel('loss')

loss_ax.legend(loc = 'upper left')

plt.show()

​

# In[14]:

​

​

def show_diff(y_predict, y_label, y_test, y_test_label, samples, y_value):

fig, predict_ax=plt.subplots()

predict_ax.plot(y_predict, 'y', label=y_label)

predict_ax.plot(y_test, 'r', label=y_test_label)

predict_ax.set_xlabel(samples)

predict_ax.set_ylabel(y_value)

predict_ax.legend(loc = 'upper left')

plt.show()

​

​

# In[15]:

​

​

#7 예측하기

pre_test_y = model.predict(x_train)

show_diff(pre_test_y, 'pred y', y_train, 'train y', 'samples', 'Y value')

y_predict = model.predict(x_test).flatten()

show_diff(y_predict, 'y_predict', y_test, 'y_test', 'samples', 'y_value')

# In[16]:

​

​

#8 저장하기

df = pd.DataFrame(y_predict)

df. insert(1, 'y_test', y_test)

df.to_csv('predict_price.csv')

​

​

# In[17]:

​

​

# compute the difference between the *predicted* house prices and the

# *actual* house prices, then compute the percentage difference and

# the absolute percentage difference

diff = y_predict - y_test

percentDiff = (diff / y_test) * 100

absPercentDiff = np.abs(percentDiff)

​

# compute the mean and standard deviation of the absolute percentage

# difference

mean = np.mean(absPercentDiff)

std = np.std(absPercentDiff)

​

# finally, show some statistics on our model

#locale.setlocale(locale.LC_ALL, "en_US.UTF-8")

print("[INFO] mean: {:.2f}%, std: {:.2f}%".format(mean, std))

[INFO] mean: 13.37%, std: 11.71%