使用回归 (regression)进行预测

1.导入模块

!pip install seaborn
import pathlib
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
import tensorflow as tf 
from tensorflow import keras
from tensorflow.keras import layers
tf.__version__

2.下载数据集

dataset_path = keras.utils.get_file('auto-img.data', 'http://archive.ics.uci.edu/ml/machine-learning-databases/auto-mpg/auto-mpg.data')
dataset_path

运行后输出：

‘C:\Users\Administrator\.keras\datasets\auto-img.data’
查看一下数据集中的内容：

column_names = ['MPG','Cylinders','Displacement','Horsepowner','Weight','Acceleration', 'Model Year','Origin']
raw_dataset = pd.read_csv(dataset_path, names=column_names,
                         na_values='?', comment='\t',
                         sep=' ',skipinitialspace=True)
dataset = raw_dataset.copy()
dataset.tail()

3.数据清洗

查看一下有没有空值：

dataset.isna().sum()

输出结果：
MPG 0
Cylinders 0
Displacement 0
Horsepowner 6
Weight 0
Acceleration 0
Model Year 0
Origin 0
dtype: int64

不管有没有空值 ，去除一下：

dataset = dataset.dropna()

用pop 命令， 把origin列做拿出来，后续作为为独热编码：

origin = dataset.pop('Origin')

给表增加列：

dataset['USA'] = (origin == 1)*1.0
dataset['Europe'] = (origin == 2)*1.0
dataset['Japan'] = (origin == 3)*1.0
dataset.tail()

在这里插入图片描述
将数据集拆分为验证集与训练集

train_dataset = dataset.sample(frac=0.8, random_state=0)
test_dataset = dataset.drop(train_dataset.index)

用seaborn 查看一下情况

sns.pairplot(train_dataset[['MPG', 'Cylinders', 'Displacement', 'Weight']], diag_kind='kde')

看一下透视统计：

train_stats = train_dataset.describe()
train_stats.pop('MPG')
train_stats = train_stats.transpose()
train_stats

把标签分离出来

train_labels = train_dataset.pop('MPG')
test_labels = test_dataset.pop('MPG')

数据归一化（每个值减去均值后再除以标准差）

def norm(x):
    return (x - train_stats['mean'])/train_stats['std']
normed_train_data = norm(train_dataset)
normed_test_data = norm(test_dataset)

4.构建模型

def build_model():
    model = keras.Sequential([
        layers.Dense(64, activation='relu', input_shape=[len(train_dataset.keys())]),
        layers.Dense(64, activation='relu'),
        layers.Dense(1)
    ])
    optimizer = tf.keras.optimizers.RMSprop(0.001)
    model.compile(loss='mse',
                 optimizer=optimizer,
                 metrics=['mae', 'mse'])
    return model

model = build_model()
model.summary()

Model: “sequential_2”

Layer (type) Output Shape Param #

dense_3 (Dense) (None, 64) 640

dense_4 (Dense) (None, 64) 4160

dense_5 (Dense) (None, 1) 65

=================================================================
Total params: 4,865
Trainable params: 4,865
Non-trainable params: 0

测试一下预测:

example_batch = normed_train_data[:10]
example_result = model.predict(example_batch)
example_result

1/1 [==============================] - 0s 311ms/step
array([[ 0.1861527 ],
[-0.03714132],
[ 0.41865367],
[ 0.08125568],
[ 0.35935453],
[ 0.14658093],
[ 0.42246234],
[ 0.07202747],
[ 0.08832908],
[ 0.41105857]], dtype=float32)

5. 开始训练

class PrintDot(keras.callbacks.Callback):
    def on_epoch_end(self, epoch, logs):
        if epoch%100 ==0:print('')
        print('.', end='')
EPOCHS = 10000

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

来看一下模型中的参数

hist= pd.DataFrame(history.history)
hist['epoch'] = history.epoch
hist.tail()

6.图形化显示

def plot_history(history):
    hist = pd.DataFrame(history.history)
    hist['epoch'] = history.epoch
    
    plt.figure()
    plt.xlabel('Epoch')
    plt.ylabel('Mean Abs Error [MGP]')
    plt.plot(hist['epoch'], hist['mae'], label='Val Error')
    plt.plot(hist['epoch'], hist['val_mae'], label='Val Error')
    plt.ylim([0,5])
    plt.legend()
    
    plt.figure()
    plt.xlabel('Epoch')
    plt.ylabel('Mean Square Error [$MGP^2$]')
    plt.plot(hist['epoch'], hist['mse'], label='Train Error')
    plt.plot(hist['epoch'], hist['val_mse'], label='Val Error')
    plt.ylim([0,20])
    plt.legend()
    plt.show()

运行一下：

plot_history(history)

再优化一下

model = build_model()
early_stop = keras.callbacks.EarlyStopping(monitor='val_loss', patience=10)
history = model.fit(normed_train_data, train_labels, epochs=EPOCHS,
                   validation_split=0.2, verbose=0, callbacks= [early_stop, PrintDot()])

plot_history(history)

loss, mae, mse = model.evaluate(normed_test_data, test_labels, verbose=2)

print("Testing set Mean Abs Error: {:5.2f} MPG".format(mae))

3/3 - 0s - loss: 6.5743 - mae: 1.8934 - mse: 6.5743 - 37ms/epoch - 12ms/step
Testing set Mean Abs Error: 1.89 MPG

用测数据预测一下MPG

test_predictions = model.predict(normed_test_data).flatten()

plt.scatter(test_labels, test_predictions)
plt.xlabel('True Values [MPG]')
plt.ylabel('Predictions [MPG]')
plt.axis('equal')
plt.axis('square')
plt.xlim([0,plt.xlim()[1]])
plt.ylim([0,plt.ylim()[1]])
_ = plt.plot([-100, 100], [-100, 100])

这看起来我们的模型预测得相当好。我们来看下误差分布。

error = test_predictions - test_labels
plt.hist(error, bins = 25)
plt.xlabel("Prediction Error [MPG]")
_ = plt.ylabel("Count")