from __future__ import division from __future__ import print_function import numpy as np import pandas as pd import matplotlib.pylab as plt import seaborn as sns import tensorflow as tf from tensorflow.python.ops import rnn, rnn_cell import tushare as ts def getData(id,start,end,num,flag): df = ts.get_hist_data(id,start,end) #df = (df-np.sum(df)/len(df))/(np.std(df)) if(flag=="true"): df = df[1:num] else: df = df[:num] df1 = np.array(df) #df2 = np.array(df.index) ##df = df.T x = [] for i in range(len(df1)): #temp = np.append(df2[i],df1[i]) temp = df1[i] newresult = [] for item in temp: newresult.append(item) x.append(newresult) x.reverse() return x def getDataR(id,start,end,num): df = ts.get_hist_data(id,start,end) df1 = np.array(df) x = [] for i in range(len(df1)): temp = df1[i] newresult = [] for item in temp: newresult.append(item) x.append(newresult) P=df['close'] #ʵ������û��end��һ������ݣ�������Ԥ��δ��һ����������ڵ����̼� templist=(P.shift(1)-P)/P templist = templist[:num] templist = np.array(templist) templist = templist.tolist() templist.reverse() tempDATA = [] for i in range(len(templist)): if((i+1)%10!=0): pass else: if(templist[i]>0): #tempDATA.append(templist[i]) tempDATA.append([1,0,0]) elif(templist[i]<=0): #tempDATA.append(templist[i]) tempDATA.append([0,1,0]) else: #tempDATA.append(templist[i]) tempDATA.append([0,0,1]) y=tempDATA return y #df_sh = ts.get_sz50s()['code'] df_sh =["600016"] fac = [] ret = [] facT = [] retT = [] predFAC = [] for ishare in df_sh: #ȡ�����260������ newfac = getData(ishare,'2008-07-22','2016-08-01',601,"true") newret = getDataR(ishare,'2008-07-22','2016-08-01',601) #fac.append(newfac) for i in range(len(newfac)): fac.append(newfac[i]) for i in range(len(newret)): ret.append(newret[i]) newfacT = getData(ishare,'2016-08-01','2017-01-19',101,"true") newretT = getDataR(ishare,'2016-08-01','2017-01-19',101) #fac.append(newfac) for i in range(len(newfacT)): facT.append(newfacT[i]) for i in range(len(newretT)): retT.append(newretT[i]) newpredFAC = getData(ishare,'2016-08-01','2017-01-20',11,"false") for i in range(len(newpredFAC)): predFAC.append(newpredFAC[i]) fac = np.array(fac) ret = np.array(ret) meanfac = np.sum(fac, axis=0)/len(fac) stdfac = np.std(fac, axis=0) fac = (fac-meanfac)/stdfac facT = np.array(facT) retT = np.array(retT) facT = (facT-meanfac)/stdfac newf = [] newfa = [] for i in range(len(fac)): if((i+1)%10!=0): newf.append(fac[i]) else: newf.append(fac[i]) newfa.append(newf) newf = [] fac = np.array(newfa) newfT = [] newfaT = [] for i in range(len(facT)): if((i+1)%10!=0): newfT.append(facT[i]) else: newfT.append(facT[i]) newfaT.append(newfT) newfT = [] facT = np.array(newfaT) predFAC = (predFAC-meanfac)/stdfac learning_rate = 0.001 batch_size = 10 print(int(fac.shape[0])) training_iters = int(fac.shape[0]/batch_size) display_step = 10 # Network Parameters n_input = 14 n_steps = 10 n_hidden = 1024 n_classes = 3 dropout = 0.8 # tf Graph input x = tf.placeholder('float',[None, n_steps, n_input]) y = tf.placeholder('float',[None, n_classes]) keep_prob = tf.placeholder(tf.float32) #dropout (keep probability) # ������� def conv2d(name, l_input, w, b): return tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(l_input, w, strides=[1, 1, 1, 1], padding='SAME'),b), name=name) # ����²������� def max_pool(name, l_input, k): return tf.nn.max_pool(l_input, ksize=[1, k, k, 1], strides=[1, k, k, 1], padding='SAME', name=name) # ��һ������ def norm(name, l_input, lsize=4): return tf.nn.lrn(l_input, lsize, bias=1.0, alpha=0.001 / 9.0, beta=0.75, name=name) # ������������ def alex_net(_X, _weights, _biases, _dropout): # ����תΪ���� _X = tf.reshape(_X, shape=[-1, 10, 14, 1]) # ����� conv1 = conv2d('conv1', _X, _weights['wc1'], _biases['bc1']) # �²����� pool1 = max_pool('pool1', conv1, k=2) # ��һ���� norm1 = norm('norm1', pool1, lsize=4) # Dropout norm1 = tf.nn.dropout(norm1, _dropout) # ��� conv2 = conv2d('conv2', norm1, _weights['wc2'], _biases['bc2']) # �²��� pool2 = max_pool('pool2', conv2, k=2) # ��һ�� norm2 = norm('norm2', pool2, lsize=4) # Dropout norm2 = tf.nn.dropout(norm2, _dropout) # ��� conv3 = conv2d('conv3', norm2, _weights['wc3'], _biases['bc3']) # �²��� pool3 = max_pool('pool3', conv3, k=2) # ��һ�� norm3 = norm('norm3', pool3, lsize=4) # Dropout norm3 = tf.nn.dropout(norm3, _dropout) # ȫ���Ӳ㣬�Ȱ�����ͼתΪ���� dense1 = tf.reshape(norm3, [-1, _weights['wd1'].get_shape().as_list()[0]]) dense1 = tf.nn.relu(tf.matmul(dense1, _weights['wd1']) + _biases['bd1'], name='fc1') # ȫ���Ӳ� dense2 = tf.nn.relu(tf.matmul(dense1, _weights['wd2']) + _biases['bd2'], name='fc2') # Relu activation # ��������� out = tf.matmul(dense2, _weights['out']) + _biases['out'] return out # �洢���е�������� weights = { 'wc1': tf.Variable(tf.random_normal([3, 3, 1, 64])), 'wc2': tf.Variable(tf.random_normal([3, 3, 64, 128])), 'wc3': tf.Variable(tf.random_normal([3, 3, 128, 256])), 'wd1': tf.Variable(tf.random_normal([1024, 1024])), 'wd2': tf.Variable(tf.random_normal([1024, 1024])), 'out': tf.Variable(tf.random_normal([1024, n_classes])) } biases = { 'bc1': tf.Variable(tf.random_normal([64])), 'bc2': tf.Variable(tf.random_normal([128])), 'bc3': tf.Variable(tf.random_normal([256])), 'bd1': tf.Variable(tf.random_normal([1024])), 'bd2': tf.Variable(tf.random_normal([1024])), 'out': tf.Variable(tf.random_normal([n_classes])) } # ����ģ�� pred = alex_net(x, weights, biases, keep_prob) # ������ʧ������ѧϰ���� cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(pred, y)) optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost) # �������� correct_pred = tf.equal(tf.argmax(pred,1), tf.argmax(y,1)) accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32)) # ��ʼ�����еĹ������ init = tf.global_variables_initializer() # ����һ��ѵ�� with tf.Session() as sess: sess.run(init) for tr in range(100): #for tr in range(3): for i in range(int(len(fac)/batch_size)): batch_x = fac[i*batch_size:(i+1)*batch_size].reshape([batch_size,n_steps,n_input]) batch_y = ret[i*batch_size:(i+1)*batch_size].reshape([batch_size,n_classes]) sess.run(optimizer,feed_dict={x:batch_x,y:batch_y,keep_prob:dropout}) if(i%50==0): print(i,'----',(int(len(fac)/batch_size))) loss, acc = sess.run([cost, accuracy], feed_dict={x: batch_x,y: batch_y, keep_prob:0.8}) print("Iter " + str(tr*batch_size) + ", Minibatch Loss= " +"{:.26f}".format(loss) + ", Training Accuracy= " +"{:.26f}".format(acc)) print("Optimization Finished!") # ������Ծ��� print("Accuracy in data set") test_data = fac[:batch_size].reshape([batch_size,n_steps,n_input]) test_label = ret[:batch_size].reshape([batch_size,n_classes]) loss, acc = sess.run([cost, accuracy], feed_dict={x: test_data,y: test_label, keep_prob:1.}) print("Accuracy= " +"{:.26f}".format(acc)) print("Accuracy out of data set") test_dataT = facT[:len(facT)].reshape([len(facT),n_steps,n_input]) test_labelT = retT[:len(facT)].reshape([len(facT),n_classes]) loss, acc = sess.run([cost, accuracy], feed_dict={x: test_dataT,y: test_labelT, keep_prob:1.}) print("Accuracy= " +"{:.26f}".format(acc)) pred_dataT = predFAC[:batch_size].reshape([1,n_steps,n_input]) pred_lable = sess.run([pred],feed_dict={x: pred_dataT, keep_prob:1.}) list_lable = pred_lable[0][0] maxindex = np.argmax(list_lable) #print("Predict_label is " + str(pred_lable[0][0])) if(maxindex==0): print("up") else: print("down") sess.close()
快疯了，也存在过拟合问题，做个笔记，以后慢慢完善
