_0_1_tf_liner

1. Pass

2. Code:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

import tensorflow as tf


def create_data():
    """创建数据，供自测学习使用"""
    num_points = 1000
    vectors_set = []
    for item in range(num_points):
        x = np.random.normal(0.00, 0.55)
        y = 0.1 * x + 0.3 + np.random.normal(0.00, 0.03)
        vectors_set.append([x, y])
    return vectors_set


def preprocessing(data):
    """源数据中提取 x,y 数据"""
    x_data = [item[0] for item in data]
    y_data = [item[1] for item in data]
    return (x_data, y_data)


def model_liner(x_data, y_data):
    # Prepare param(weight,bias)
    w = tf.Variable(tf.random_uniform([1], -1.0, 1.0), name='weight')
    b = tf.Variable(tf.zeros([1]), name='bias')

    # Decision fun(Y=W*X.T+b)
    y = w * x_data + b

    # Loss fun(loss=tf.reduce_mean(tf.square(y-y_data)))
    loss = tf.reduce_mean(tf.square(y - y_data), name='loss')

    # Optimizer algorithm(GradientDescentOptimizer)
    # TODO learning_rate 作为配置文件中 变动参数，依据训练结果来进行调整
    optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.3)

    # Create Training fun(train = optimizer.minimize(loss,name='train'))
    train = optimizer.minimize(loss, name='train')

    epoch = 20

    '''开启训练通道'''
    sess = tf.Session()
    sess.run(tf.global_variables_initializer())
    print('w=', sess.run(w), '\tb=', sess.run(b), '\tloss:', sess.run(loss))  # TODO test
    for item in range(epoch):
        sess.run(train)
        print('w=', sess.run(w), '\tb=', sess.run(b), '\tloss:', sess.run(loss))  # TODO test

    w, b = sess.run(w), sess.run(b)
    sess.close()

    return (w, b)


def show_img(x_data, y_data, w, b):
    """展示 原始数据，拟合成的直线"""
    plt.scatter(x_data, y_data, c='r')
    plt.plot(x_data, w * x_data + b, c='b')
    plt.show()


def process():
    data = create_data()
    x_data, y_data = preprocessing(data)
    w, b = model_liner(x_data, y_data)
    print('\n', w, b, '\n')
    show_img(x_data, y_data, w, b)


if __name__ == '__main__':
    process()

3. 代码更新...

4. 备注