企业官网建设流程全解析

先是data_processing文件：

import pandas as pd from urllib.request import urlretrieve def load_data(download=True): # download data from : http://archive.ics.uci.edu/ml/datasets/Car+Evaluation if download: data_path, _ = urlretrieve("http://archive.ics.uci.edu/ml/machine-learning-databases/car/car.data", "car.csv") print("Downloaded to car.csv") # use pandas to view the data structure col_names = ["buying", "maint", "doors", "persons", "lug_boot", "safety", "class"] data = pd.read_csv("car.csv", names=col_names) return data def convert2onehot(data): # covert data to onehot representation return pd.get_dummies(data, prefix=data.columns) if __name__ == "__main__": data = load_data(download=True) new_data = convert2onehot(data) print(data.head()) print("\nNum of data: ", len(data), "\n") # 1728 # view data values for name in data.keys(): print(name, pd.unique(data[name])) print("\n", new_data.head(2)) new_data.to_csv("car_onehot.csv", index=False)

1.pd.read_csv
功能：给数据添加一行列索引，也就是横在数据上面的表头
返回一个诡异的数据类型：
data: <class ‘pandas.core.frame.DataFrame’>

col_names = ["buying", "maint", "doors", "persons", "lug_boot", "safety", "class"] data = pd.read_csv("car.csv", names=col_names)

2.get_dummies(data, prefix=data.columns)
pandas的功能真的很强大，可以直接以data.columns为标准生成独热码形式
https://blog.csdn.net/qq_35290785/article/details/91415240

return pd.get_dummies(data, prefix=data.columns)

3.data.head()
读取前五行数据
https://blog.csdn.net/qq_18649781/article/details/89033749

print(data.head())

4.pd.unique
用法自己琢磨把…

for name in data.keys(): print(name, pd.unique(data[name]))

5.to_csv
dt.to_csv() #默认dt是DataFrame的一个实例，
index=False表示不保留行索引，也就是不保留最左侧的竖列索引

new_data.to_csv("car_onehot.csv", index=False)

综上所述，data_processing代码完成了one-hot的csv文件的建立

接下来 我们来看model文件

import numpy as np import tensorflow as tf import matplotlib.pyplot as plt import data_processing data = data_processing.load_data(download=True) new_data = data_processing.convert2onehot(data) # prepare training data new_data = new_data.values.astype(np.float32) # change to numpy array and float32 np.random.shuffle(new_data) sep = int(0.7*len(new_data)) train_data = new_data[:sep] # training data (70%) test_data = new_data[sep:] # test data (30%) # build network tf_input = tf.placeholder(tf.float32, [None, 25], "input") tfx = tf_input[:, :21] tfy = tf_input[:, 21:] l1 = tf.layers.dense(tfx, 128, tf.nn.relu, name="l1") l2 = tf.layers.dense(l1, 128, tf.nn.relu, name="l2") out = tf.layers.dense(l2, 4, name="l3") prediction = tf.nn.softmax(out, name="pred") loss = tf.losses.softmax_cross_entropy(onehot_labels=tfy, logits=out) accuracy = tf.metrics.accuracy( # return (acc, update_op), and create 2 local variables labels=tf.argmax(tfy, axis=1), predictions=tf.argmax(out, axis=1))[1] opt = tf.train.GradientDescentOptimizer(learning_rate=0.1) train_op = opt.minimize(loss) sess = tf.Session() sess.run(tf.group(tf.global_variables_initializer(), tf.local_variables_initializer())) # training plt.ion() fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8, 4)) accuracies, steps = [], [] for t in range(4000): # training batch_index = np.random.randint(len(train_data), size=32) sess.run(train_op, {tf_input: train_data[batch_index]}) if t % 50 == 0: # testing acc_, pred_, loss_ = sess.run([accuracy, prediction, loss], {tf_input: test_data}) accuracies.append(acc_) steps.append(t) print("Step: %i" % t,"| Accurate: %.2f" % acc_,"| Loss: %.2f" % loss_,) # visualize testing ax1.cla() for c in range(4): bp = ax1.bar(c+0.1, height=sum((np.argmax(pred_, axis=1) == c)), width=0.2, color='red') bt = ax1.bar(c-0.1, height=sum((np.argmax(test_data[:, 21:], axis=1) == c)), width=0.2, color='blue') ax1.set_xticks(range(4), ["accepted", "good", "unaccepted", "very good"]) ax1.legend(handles=[bp, bt], labels=["prediction", "target"]) ax1.set_ylim((0, 400)) ax2.cla() ax2.plot(steps, accuracies, label="accuracy") ax2.set_ylim(ymax=1) ax2.set_ylabel("accuracy") plt.pause(0.01) plt.ioff() plt.show()

1.tf.nn.softmax
将所选择的一维数组归一化，也就是把每一个元素都变成0-1之间的数字

2.tf.metrics.accuracy
tf.metrics.accuracy返回两个值，accuracy为到上一个batch为止的准确度，update_op为更新本批次后的准确度。
https://blog.csdn.net/lyb3b3b/article/details/83047148
因为有局部变量，所以要初始化局部变量，sess.run(tf.local_variables_initializer())
这就是为什么最后有初始化局部变量的原因

3.plt.ion()
打开交互模式：若想动态显示图像，则需要使用交互（interactive）模式。
https://www.cnblogs.com/wmy-ncut/p/10172601.html

4.plt.subplots
前两个数字是子图的行列数
figsize=(8, 4)估计就是确定figure的大小

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8, 4))

5.numpy.random.randint(low, high=None, size=None, dtype=‘l’)
函数的作用是，返回一个随机整型数，范围从低（包括）到高（不包括），即[low, high)。
如果没有写参数high的值，则返回[0,low)的值。
size: int or tuple of ints(可选)
输出随机数的尺寸，比如size = (m * n* k)则输出同规模即m * n* k个随机数。默认是None的，仅仅返回满足要求的单一随机数。
https://blog.csdn.net/u011851421/article/details/83544853
即：整数，生成随机元素的个数或者元组，数组的行和列
也就是这里生成的是array([1,5,6,8,4,3…]) 共32个元素

batch_index = np.random.randint(len(train_data), size=32)

6.现在不用feed_dict={}这种形式了
用：

sess.run(train_op, {tf_input: train_data[batch_index]})

7.ax1.cla()
plt.cla() # 清除axes（轴），即清楚当前 figure 中的活动的axes，但其他axes保持不变。
暂时存疑

8.ax1.bar
https://blog.csdn.net/liangzuojiayi/article/details/78187704

1. left：x轴的位置序列，一般采用arange函数产生一个序列；
2. height：y轴的数值序列，也就是柱形图的高度，一般就是我们需要展示的数据；
3. alpha：透明度
4. width：为柱形图的宽度，一般这是为0.8即可；
5. color或facecolor：柱形图填充的颜色；
6. edgecolor：图形边缘颜色
7. label：解释每个图像代表的含义
8. linewidth or linewidths or lw：边缘or线的宽度

bp = ax1.bar(c+0.1, height=sum((np.argmax(pred_, axis=1) == c)), width=0.2, color='red')

后面的先不写了，matplotlib有点麻烦

补充：
1.dataframe格式建立：
https://blog.csdn.net/qq_39161737/article/details/78866399
https://blog.csdn.net/weixin_40240670/article/details/80506402

加入了save，并且单独摘2个测试 data_processing:

import pandas as pd from urllib.request import urlretrieve import numpy as np def load_data(download=True): # download data from : http://archive.ics.uci.edu/ml/datasets/Car+Evaluation if download: data_path, _ = urlretrieve("http://archive.ics.uci.edu/ml/machine-learning-databases/car/car.data", "car.csv") print("Downloaded to car.csv") # use pandas to view the data structure col_names = ["buying", "maint", "doors", "persons", "lug_boot", "safety", "class"] data = pd.read_csv("car.csv", names=col_names) data_test = pd.read_csv("car_test.csv", names=col_names) # print(data_test) #<class 'pandas.core.frame.DataFrame'> return data, data_test def convert2onehot(data, train): # covert data to onehot representation if train == True: print(type(pd.get_dummies(data, prefix=data.columns))) return pd.get_dummies(data, prefix=data.columns) else: x = [[0,0,0,1,0,0,0,1,1,0,0,0,1,0,0,0,0,1,0,1,0,0,0,1,0],[0,0,0,1,0,0,0,1,1,0,0,0,1,0,0,0,0,1,0,0,1,0,0,1,0]] return pd.DataFrame(x) if __name__ == "__main__": data, data_test= load_data(download=True) new_data = convert2onehot(data, train=True) new_data_test = convert2onehot(data_test, train=False) # new_data_test = new_data_test.values.astype(np.float32) # print("自己建立的是：", new_data_test) # print("new:", new_data) print(data.head()) print("\nNum of data: ", len(data), "\n") # 1728 也就是1728行 # view data values for name in data.keys(): print(name, pd.unique(data[name])) print("\n", new_data.head(2)) new_data.to_csv("car_onehot.csv", index=False) # new_data_test.to_csv("car_onehot_test.csv", index=False)

model:

import numpy as np import tensorflow as tf import matplotlib.pyplot as plt import data_processing import os tf.app.flags.DEFINE_integer("is_train", 2, "指定是否是训练模型，还是拿数据去预测") FLAGS = tf.app.flags.FLAGS data, data_test = data_processing.load_data(download=True) new_data = data_processing.convert2onehot(data, train=True) new_data_test = data_processing.convert2onehot(data, train=False) # prepare training data new_data = new_data.values.astype(np.float32) # change to numpy array and float32 转换为numpy数组形式并规定为浮点数 new_data_test = new_data_test.values.astype(np.float32) #print("打印：", new_data) np.random.shuffle(new_data) sep = int(0.7*len(new_data)) train_data = new_data[:sep] # training data (70%) test_data = new_data[sep:] # test data (30%) realtest_data = new_data_test[::] np.random.shuffle(new_data_test) # build network tf_input = tf.placeholder(tf.float32, [None, 25], "input") #其中最后4个是汽车状态评价 tfx = tf_input[:, :21] tfy = tf_input[:, 21:] # tfx_test = tf_input[:, :21] # tfy_test = tf_input[:, 21:] l1 = tf.layers.dense(tfx, 128, tf.nn.relu, name="l1") l2 = tf.layers.dense(l1, 128, tf.nn.relu, name="l2") out = tf.layers.dense(l2, 4, name="l3") prediction = tf.nn.softmax(out, name="pred") # print("prediction:", prediction) loss = tf.losses.softmax_cross_entropy(onehot_labels=tfy, logits=out) accuracy = tf.metrics.accuracy( # return (acc, update_op), and create 2 local variables labels=tf.argmax(tfy, axis=1), predictions=tf.argmax(out, axis=1))[1] opt = tf.train.GradientDescentOptimizer(learning_rate=0.1) train_op = opt.minimize(loss) #进行模型保存 # (2)收集要显示的变量 # 先收集损失和准确率 tf.summary.scalar("losses", loss) tf.summary.scalar("accuracy", accuracy) # (3)合并所有变量op merged = tf.summary.merge_all() # 创建模型保存与加载 saver = tf.train.Saver() with tf.Session() as sess: sess.run(tf.group(tf.global_variables_initializer(), tf.local_variables_initializer())) # (1)创建一个events文件实例 file_writer = tf.summary.FileWriter("./tmp/summary2/", graph=sess.graph) # 加载模型 if os.path.exists("./tmp/modelckpt2/checkpoint"): saver.restore(sess, "./tmp/modelckpt2/cnn_model") # 注意modelckpt2这个文件夹要自己建立 # 也就是说 模型保存和加载的时候 也就是saver.save或saver.restore的路径需要自己建立 否则会蓝屏 # 但是创建envents实例化的路径可以不用自己建立 # training+test1 if FLAGS.is_train == 1: accuracies, steps = [], [] for t in range(4000): # training batch_index = np.random.randint(len(train_data), size=32) #生成32个随机数，返回array数组 sess.run(train_op, {tf_input: train_data[batch_index]}) #注意feeddict可以不用了 # 运行合变量op，写入事件文件当中 summary = sess.run(merged, {tf_input: train_data[batch_index]}) file_writer.add_summary(summary, t) if t % 2 == 0: saver.save(sess, "./tmp/modelckpt2/cnn_model") if t % 50 == 0: # testing acc_, pred_, loss_ = sess.run([accuracy, prediction, loss], {tf_input: test_data}) accuracies.append(acc_) steps.append(t) print("Step: %i" % t,"| Accurate: %.2f" % acc_,"| Loss: %.2f" % loss_,) # test2 else: for t in range(2): # training batch_index = np.random.randint(len(realtest_data), size=1) # sess.run(train_op, {tf_input_test: test_real[batch_index]}) acc_, pred_, loss_ = sess.run([accuracy, prediction, loss], {tf_input: realtest_data[batch_index]}) print("真实值为：", sess.run(tf.argmax(realtest_data[batch_index][:, 21:], axis=1))) print("预测值为：", sess.run(tf.argmax(pred_, axis=1)))