我试图对由Keras模型预测结果输出的numpy数组取整。但是,在执行numpy.round / numpy.around之后,没有任何变化。
此处的最终目标是,如果数组小于或等于0.50,则将数组舍入为0;如果数组大于0.50,则将数组向上舍入。
代码在这里:
from keras.models import load_model
import numpy
model = load_model('tried.h5')
data = numpy.loadtxt("AppData\Roaming\MetaQuotes\Terminal\94DDB309C90B408373EFC53AC730F336\MQL4\Files\indicatorout.csv", delimiter=",")
data = numpy.array([data])
print(data)
outdata = model.predict(data)
print(outdata)
numpy.around(outdata, 0)
print(outdata)
numpy.savetxt("AppData\Roaming\MetaQuotes\Terminal\94DDB309C90B408373EFC53AC730F336\MQL4\Files\modelout.txt", outdata)
日志也在这里:
Using TensorFlow backend.
[[1.19539070e+01 1.72686310e+01 2.24426384e+01 1.82771435e+01
2.23788052e+01 1.62105408e+01 1.44595184e+01 1.90179043e+01
1.71749554e+01 1.69194088e+01 1.89911938e+01 1.76701393e+01
5.19613740e-01 5.38522415e+01 9.64037247e+01 1.73570000e-04
4.35710000e-04 9.55710000e-04]]
[[0.4215713]]
[[0.4215713]]
任何帮助将不胜感激,谢谢。
答案 0 :(得分:1)
我假设您希望数组中的元素四舍五入到n
个小数位。下面是这样做的示意图:
# sample array to work with
In [21]: arr = np.random.randn(4)
In [22]: arr
Out[22]: array([-0.94817409, -1.61453252, 0.16566428, -0.53507549])
# round to 3 decimal places; note that `arr` is still unaffected.
In [23]: arr.round(decimals=3)
Out[23]: array([-0.948, -1.615, 0.166, -0.535])
# if you want to round it to nearest integer
In [24]: arr_rint = np.rint(arr)
In [25]: arr_rint
Out[25]: array([-1., -2., 0., -1.])
要使小数舍入就位,请按如下所示指定out=
参数:
In [26]: arr.round(decimals=3, out=arr)
Out[26]: array([-0.948, -1.615, 0.166, -0.535])