def ai_mul_var_tst(mx, df_train, df_test, nepochs=200, nsize=128, ky0=5):
x_train, y_train = df_train['x'].values, df_train['y'].values
x_test, y_test = df_test['x'].values, df_test['y'].values
#
mx.fit(x_train, y_train, epochs=nepochs, batch_size=nsize)
#
y_pred = mx.predict(x_test)
df_test['y_pred'] = zdat.ds4x(y_pred, df_test.index, True)
dacc, _ = ai_acc_xed2x(df_test.y, df_test['y_pred'], ky0, False)
#
return dacc
mx.summary()
plot_model(mx, to_file='tmp/lstm010mx.png')
#4 模型训练
print('\n#4 模型训练 fit')
tbCallBack = keras.callbacks.TensorBoard(log_dir=rlog,
write_graph=True,
write_images=True)
tn0 = arrow.now()
mx.fit(x_train,
y_train,
epochs=2000,
batch_size=512 * 24,
callbacks=[tbCallBack])
tn = zt.timNSec('', tn0, True)
mx.save('tmp/lstm010mx.dat')
#5 利用模型进行预测 predict
print('\n#5 模型预测 predict')
tn0 = arrow.now()
y_pred = mx.predict(x_test)
tn = zt.timNSec('', tn0, True)
df_test['y_pred'] = zdat.ds4x(y_pred, df_test.index, True)
df_test.to_csv('tmp/df_lstm010mx.csv', index=False)
#6
print('\n#6 acc准确度分析')
print('\nky0=10')
df = df_test
dacc, dfx, a10 = ztq.ai_acc_xed2ext(df.y, df.y_pred, ky0=10, fgDebug=True)
print(type(X))
#5
print('\n#5,构建线性回归神经网络模型')
input_ = tflearn.input_data(shape=[None])
linear = tflearn.single_unit(input_)
regression = tflearn.regression(linear,
optimizer='sgd',
loss='mean_square',
metric='R2',
learning_rate=0.01)
m = tflearn.DNN(regression, tensorboard_dir=rlog)
#6
print('\n#6,开始训练模型')
m.fit(X, Y, n_epoch=100, show_metric=True, snapshot_epoch=False)
#7
print('\n#7,根据模型，进行预测')
X2 = df2['close'].values
Y2 = m.predict(X2)
#
ds2y = zdat.ds4x(Y2, df2.index)
df2['open2'] = ds2y
#
print(df2.tail())
df2.to_csv('tmp/df2.csv')
#9
print('\n#9,ok')
#
#6 利用模型进行预测 predict
print('\n#6 模型预测 predict')
y_pred = model.predict(x_test)
print(y_pred)
print('type(y_pred):', type(y_pred))
#7 整理预测数据
print('\n#7 整理预测数据 ')
print('\n acc.xed')
y2x = y_pred.flatten()[:]
print('\n y2x')
print(y2x)
print('type(y2x):', type(y2x))
ds2y = zdat.ds4x(y2x, df_test.index)
df_test['y_pred'] = ds2y
print('\ndf_test')
print(df_test.tail(9))
print('\n acc.xed')
dacc, df2, a10 = ztq.ai_acc_xed2ext(df_test.y,
df_test.y_pred,
ky0=5,
fgDebug=True)
#
#
#8 draw
print('\n#8 绘制图形')
v1 = [df_test.x, df_test.y, 300, 'blue', 0.2]
v2 = [df_test.x, df_test.y_pred, 50, 'red', 0.6]
net = tflearn.fully_connected(net, 3, activation='softmax')
net = tflearn.regression(net)
#
m = tflearn.DNN(net, tensorboard_dir=rlog)
#
#6
print('\n#6,开始训练模型')
m.fit(X, y1, n_epoch=100, show_metric=True)
#7
print('\n#7,根据模型，进行预测')
X2 = df2[zsys.ohlcLst].values
Y2 = m.predict(X2)
#
y2v = map(np.argmax, Y2)
ds2y = zdat.ds4x(y2v, df2.index)
df2['ktype2'] = ds2y
#
print(df2.tail())
df2.to_csv('tmp/df2.csv')
#
#8
print('\n#8,计算预测结果')
df5 = pd.DataFrame()
df5['y_test'] = df2['ktype']
df5['y_pred'] = df2['ktype2']
acc, df5x = ztq.ai_acc_xed2x(df5['y_test'], df5['y_pred'], ky0=0.5)
#
print('\nacc,', acc)
print(df5.tail())
df5.to_csv('tmp/df5.csv')
