本文旨在使用神经网络，不一定网络最佳，参数调整过程还有待商榷
第一步：分析数据集
本次采取并非已有的外汇返佣数据集，而是用文本形式给出。除最后一列以外其他列均为因素，暂不采取降维方式，直接作为影响因素加入神经网络训练。
第二步:处理数据
1.获取数据
#导入islice的目的是为了有表头存在便于剪切 from itertools import islice data_X = [] data_Y = [] with open ('data/boston_house_prices.csv') as f: for line in islice(f, 0, None): line = line.split(',') data_X.append(line[:-1]) data_Y.append(line[-1:]) # 转换为nparray data_X = np.array(data_X, dtype='float32') data_Y = np.array(data_Y, dtype='float32') print('data shape', data_X.shape, data_Y.shape) print('data_x shape[1]', data_X.shape[1])
2.数据归一化处理
# 归一化 for i in range(data_X.shape[1]): _min = np.min(data_X[:, i]) # 每一列的最小值 _max = np.max(data_X[:, i]) # 每一列的最大值 data_X[:, i] = (data_X[:, i] - _min) / (_max - _min) # 归一化到0-1之间
3.数据存入矩阵和改变维度
X_train, X_test, y_train, y_test = train_test_split(data_X, data_Y, test_size=0.5, random_state=0) X_train = scale(X_train) X_test = scale(X_test) y_train = scale(y_train.reshape((-1, 1))) y_test = scale(y_test.reshape((-1, 1)))
第三步：网络处理
1.定义网络，抽取网络，便于灵活性
# 使网络更灵活 def add_layer(inputs,input_size,output_size,activation_function=None): with tf.variable_scope("Weights"): Weights = tf.Variable(tf.random_normal(shape=[input_size, output_size]), name="weights") with tf.variable_scope("biases"): biases = tf.Variable(tf.zeros(shape=[1,output_size]) + 0.1,name="biases") with tf.name_scope("Wx_plus_b"): Wx_plus_b = tf.matmul(inputs,Weights) + biases with tf.name_scope("dropout"): Wx_plus_b = tf.nn.dropout(Wx_plus_b,keep_prob=keep_prob_s) if activation_function is None: return Wx_plus_b else: with tf.name_scope("activation_function"): return activation_function(Wx_plus_b)
2.参数设置
# 参数设置 xs = tf.placeholder(shape=[None,X_train.shape[1]],dtype=tf.float32,name="inputs") ys = tf.placeholder(shape=[None,1],dtype=tf.float32,name="y_true") keep_prob_s = tf.placeholder(dtype=tf.float32)
3.网络层
with tf.name_scope("layer_1"): l1 = add_layer(xs,13,10,activation_function=tf.nn.relu) # with tf.name_scope("layer_2"): # l2 = add_layer(l1,6,10,activation_function=tf.nn.relu) with tf.name_scope("y_pred"): pred = add_layer(l1,10,1)
4.定义代价函数
pred = tf.add(pred,0,name='pred') with tf.name_scope("loss"): loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - pred),reduction_indices=[1])) # mse tf.summary.scalar("loss",tensor=loss) with tf.name_scope("train"): train_op = tf.train.AdamOptimizer(learning_rate=0.0001).minimize(loss)
5.可视化
# 可视化 # draw pics fig = plt.figure() ax = fig.add_subplot(1,1,1) ax.plot(range(50),y_train[0:50],'b') #展示前50个数据 ax.set_ylim([-2,5]) plt.ion() plt.show()
6.参数设置
# parameters keep_prob = 0.5 # 防止过拟合，取值一般在0.5到0.8。 ITER = 15000 # 训练次数
7.定义训练方法
# 训练定义 def get_batch(image, label, batch_size, now_batch, total_batch): if now_batch < total_batch: image_batch = image[now_batch*batch_size:(now_batch+1)*batch_size] label_batch = label[now_batch*batch_size:(now_batch+1)*batch_size] else: image_batch = image[now_batch*batch_size:] label_batch = label[now_batch*batch_size:] return image_batch, label_batch def fit(X, y, ax, n, keep_prob): init = tf.global_variables_initializer() feed_dict_train = {ys: y, xs: X, keep_prob_s: keep_prob} with tf.Session() as sess: saver = tf.train.Saver(tf.global_variables(), max_to_keep=15) merged = tf.summary.merge_all() writer = tf.summary.FileWriter(logdir="nn_boston_log", graph=sess.graph) #写tensorbord sess.run(init) for i in range(n): _loss, _ = sess.run([loss, train_op], feed_dict=feed_dict_train) if i % 100 == 0: print("epoch:%d\tloss:%.5f" % (i, _loss)) y_pred = sess.run(pred, feed_dict=feed_dict_train) rs = sess.run(merged, feed_dict=feed_dict_train) writer.add_summary(summary=rs, global_step=i) # 写tensorbord saver.save(sess=sess, save_path="nn_boston_model/nn_boston.model", global_step=i) # 保存模型 try: ax.lines.remove(lines[0]) except: pass lines = ax.plot(range(50), y_pred[0:50], 'r--') plt.pause(1) saver.save(sess=sess, save_path="nn_boston_model/nn_boston.model", global_step=n) # 保存模型
8.定义测试函数
def test(X, y, keep_prob): with tf.Session() as sess: saver = tf.train.Saver(tf.global_variables(), max_to_keep=15) batch_size = 1 saver.restore(sess, tf.train.latest_checkpoint("nn_boston_model/")) # 加载变量值 print('finish loading model!') # test test_total_batch = int(len(y_test) / batch_size) test_accuracy_list = [] test_loss_list = [] y_truly = [] y_pre = [] x_test_batch_list = [] for j in range(test_total_batch): x_test_batch, y_test_batch = get_batch(X, y, batch_size, j, test_total_batch) test_accuracy, test_loss = sess.run([pred, loss], feed_dict={xs: x_test_batch, ys: y_test_batch, keep_prob_s: keep_prob}) test_loss_list.append(test_loss) y_truly.append(y_test_batch[0][0]) y_pre.append(test_accuracy[0][0]) x_test_batch_list.append(x_test_batch) print('test_loss:' + str(np.mean(test_loss_list))) print(y_pre) plt.plot(range(test_total_batch),y_truly, 'r') plt.plot(range(test_total_batch),y_pre, 'b') plt.show()
9.训练并测试
fit(X=X_train, y=y_train, n=ITER, keep_prob=keep_prob, ax=ax)
fit(X=X_test, y=y_test, keep_prob=1)