我们用 0.001 的学习速率训练该模型 50 次，即把整个训练数据集的系数曝光 50 次。运行一个 epoch 系统就将该次循环中的和方差（sum squared error）和以及最终系数集合 print 一次：

>epoch=45, lrate=0.001, error=2.650

>epoch=46, lrate=0.001, error=2.627

>epoch=47, lrate=0.001, error=2.607

>epoch=48, lrate=0.001, error=2.589

>epoch=49, lrate=0.001, error=2.573

[0.22998234937311363, 0.8017220304137576]

可以看到误差是如何在历次 epoch 中持续降低的。或许我们可以增加训练次数（epoch）或者每个 epoch 中的系数总量（调高学习速率）。

尝试一下看你能得到什么结果。

现在，我们将这个算法用到实际的数据当中。

3. 葡萄酒品质预测

我们将使用随机阶梯下降的方法为葡萄酒品质数据集训练一个线性回归模型。本示例假定一个名为 winequality—white.csv 的 csv 文件副本已经存在于当前工作目录。

首先加载该数据集，将字符串转换成数字，并将输出列从字符串转换成数值 0 和 1. 这个过程是通过辅助函数 load_csv()、str_column_to_float() 以及 dataset_minmax() 和 normalize_dataset() 来分别实现的。

我们将通过 K 次交叉验证来预估得到的学习模型在未知数据上的表现。这就意味着我们将创建并评估 K 个模型并预估这 K 个模型的平均误差。辅助函数 cross_validation_split()、rmse_metric() 和 evaluate_algorithm() 用于求导根均方差以及评估每一个生成的模型。

我们用之前创建的函数 predict()、coefficients_sgd() 以及 linear_regression_sgd() 来训练模型。完整代码如下：

# Linear Regression With Stochastic Gradient Descent for Wine Quality

from random import seed

from random import randrange

from csv import reader

from math import sqrt

# Load a CSV file

def load_csv(filename):

dataset = list()

with open(filename, 'r') as file:

csv_reader = reader(file)

for row in csv_reader:

if not row:

continue

dataset.append(row)

return dataset

# Convert string column to float

def str_column_to_float(dataset, column):

for row in dataset:

row[column] = float(row[column].strip())

# Find the min and max values for each column

def dataset_minmax(dataset):

minmax = list()

for i in range(len(dataset[0])):

col_values = [row[i] for row in dataset]

value_min = min(col_values)

value_max = max(col_values)

minmax.append([value_min, value_max])

return minmax

# Rescale dataset columns to the range 0-1

def normalize_dataset(dataset, minmax):

for row in dataset:

for i in range(len(row)):

row[i] = (row[i] - minmax[i][0]) / (minmax[i][1] - minmax[i][0])

# Split a dataset into k folds

def cross_validation_split(dataset, n_folds):

dataset_split = list()

dataset_copy = list(dataset)

fold_size = len(dataset) / n_folds

for i in range(n_folds):

fold = list()

while len(fold) < fold_size:

index = randrange(len(dataset_copy))

fold.append(dataset_copy.pop(index))

dataset_split.append(fold)

return dataset_split

# Calculate root mean squared error

def rmse_metric(actual, predicted):

sum_error = 0.0

for i in range(len(actual)):

prediction_error = predicted[i] - actual[i]

sum_error += (prediction_error ** 2)

mean_error = sum_error / float(len(actual))

return sqrt(mean_error)

# Evaluate an algorithm using a cross validation split

def evaluate_algorithm(dataset, algorithm, n_folds, *args):

folds = cross_validation_split(dataset, n_folds)

scores = list()

for fold in folds:

train_set = list(folds)

train_set.remove(fold)

train_set = sum(train_set, [])

test_set = list()

for row in fold:

row_copy = list(row)

test_set.append(row_copy)

row_copy[-1] = None

predicted = algorithm(train_set, test_set, *args)

actual = [row[-1] for row in fold]

rmse = rmse_metric(actual, predicted)

scores.append(rmse)

return scores

# Make a prediction with coefficients

def predict(row, coefficients):

yhat = coefficients[0]

for i in range(len(row)-1):

yhat += coefficients[i + 1] * row[i]

return yhat

# Estimate linear regression coefficients using stochastic gradient descent

def coefficients_sgd(train, l_rate, n_epoch):

coef = [0.0 for i in range(len(train[0]))]

for epoch in range(n_epoch):

for row in train:

yhat = predict(row, coef)

error = yhat - row[-1]

coef[0] = coef[0] - l_rate * error

for i in range(len(row)-1):

coef[i + 1] = coef[i + 1] - l_rate * error * row[i]

# print(l_rate, n_epoch, error)

return coef

# Linear Regression Algorithm With Stochastic Gradient Descent

def linear_regression_sgd(train, test, l_rate, n_epoch):

predictions = list()

coef = coefficients_sgd(train, l_rate, n_epoch)

for row in test:

yhat = predict(row, coef)

predictions.append(yhat)

return(predictions)

# Linear Regression on wine quality dataset

seed(1)

# load and prepare data

filename = 'winequality-white.csv'

dataset = load_csv(filename)

for i in range(len(dataset[0])):

str_column_to_float(dataset, i)

# normalize

minmax = dataset_minmax(dataset)

normalize_dataset(dataset, minmax)

# evaluate algorithm

n_folds = 5

l_rate = 0.01

n_epoch = 50

scores = evaluate_algorithm(dataset, linear_regression_sgd, n_folds, l_rate, n_epoch)

print('Scores: %s' % scores)

print('Mean RMSE: %.3f' % (sum(scores)/float(len(scores))))