[tensorflow2.0]08.wide_and_deep

最新推荐文章于 2020-08-12 23:24:45 发布

原创最新推荐文章于 2020-08-12 23:24:45 发布 · 451 阅读

0 ·

CC 4.0 BY-SA版权

文章标签：

#tensorflow

TensorFlow_notebook 专栏收录该内容

35 篇文章

订阅专栏

本文通过Keras的函数式API和子类API实现加州房价预测模型，详细介绍了数据预处理、模型构建、训练及评估过程。使用了California Housing数据集，并通过标准化、模型训练和验证曲线绘制展示了模型的学习效果。

import matplotlib as mpl
import matplotlib.pyplot as plt

import numpy as np
import pandas as pd
import sklearn
import os
import sys
import time
import tensorflow as tf
import pprint

from tensorflow import keras

print('Tensorflows Version:{}'.format(tf.__version__))
# print('Is gpu available:{}'.format(tf.test.is_gpu_available()))
print(sys.version_info)
for module in mpl, np, pd, sklearn, tf, keras:
    print(module.__name__, module.__version__)


from sklearn.datasets import fetch_california_housing
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler

housing = fetch_california_housing()
x_train_all, x_test, y_train_all, y_test = train_test_split(
    housing.data, housing.target, random_state=7, test_size=0.25)
x_train, x_vaild, y_train, y_vaild = train_test_split(
    x_train_all, y_train_all, random_state=7, test_size=0.25)

scaler = StandardScaler()
x_train_scaler = scaler.fit_transform(x_train)
x_vaild_scaler = scaler.transform(x_vaild)
x_test_scaler = scaler.transform(x_test)

'''
# 函数API
input = keras.layers.Input(shape=x_train_scaler.shape[1:])
hidden1 = keras.layers.Dense(30, activation='relu')(input)
hidden2 = keras.layers.Dense(30, activation='relu')(hidden1)
concat  = keras.layers.concatenate([input, hidden2])
output  = keras.layers.Dense(1)(concat)

#固化模型
model = keras.models.Model(inputs=[input], outputs=[output])
'''

# 子类API
class WideDeepModel(keras.models.Model):
    def __init__(self):
        super(WideDeepModel, self).__init__()
        '定义模型层次'
        self.hidden1_layer = keras.layers.Dense(30, activation='relu')
        self.hidden2_layer = keras.layers.Dense(30, activation='relu')
        self.output_layer  = keras.layers.Dense(1)

    def call(self, inputs, training=None, mask=None):
        '完成模型的正向计算'
        hidden1 = self.hidden1_layer(inputs)
        hidden2 = self.hidden2_layer(hidden1)
        concat  = keras.layers.concatenate([inputs, hidden2])
        return self.output_layer(concat)

model = WideDeepModel()
model.build(input_shape=(None, 8))

model.compile(optimizer='adam',
              loss=keras.losses.mean_absolute_error)

callbacks = [keras.callbacks.EarlyStopping(patience=5, min_delta=1e-3)]

history = model.fit(x_train_scaler, y_train,
                    epochs=100,
                    validation_data=(x_vaild_scaler, y_vaild),
                    callbacks=callbacks)

def plot_learning_curves(history):
    pd.DataFrame(history.history).plot(figsize=(8,5))
    plt.grid(True)
    plt.gca().set_ylim(0,1)
    plt.show()

plot_learning_curves(history)

print('model.evaluate==================')
model.evaluate(x_test_scaler, y_test)