1

我正在编写用于查找 K 近邻的代码,并且需要使用 numPy Python 包进行计算。我附上下面的功能代码:

#Import packages
import numpy as np

def knn(Q,R,k,classvar):
#Function that does k-NN and selects the k Nearest Neighbours
#Input parameters
#Q array of Query points, R array of reference points
#k is the no of closest points in k-NN
#classvar is binary class variable either 1 or 0


#for loop over all Query points
#calculate distance from ith Query point to all reference points
        dist = np.apply_along_axis(np.linalg.norm, 1, R-Q)
#sort distances and find indexes
        index=np.argsort(dist)
#pick the kth closest indexes
        knearest=index[1:k]
#find the count of 1's and 0's in the binary class variable
#if count of 1 is more than count of 0 predict class 1 else class 0
        if sum(e==0) > sum(e==1):
           preclassQ=0;
        else:
           preclassQ=1;
        return preclassQ

首先我想知道,是否有必要包含声明

from numpy import *

在函数定义中,我们将在 iPython 命令行中给出它吗?

我将此代码保存为knn.py,然后在命令行中将其导入并调用该函数,如下所示:

In [58]: import knn

In [59]: knn.knn(b,a,1,[(0),( 0),( 1),( 1)])

但是该函数没有执行,我收到如下错误!:

---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)

/Project/<ipython console> in <module>()

/Project/knn.py in knn(Q, R, k, classvar)
     14         dist = np.apply_along_axis(np.linalg.norm, 1, R-Q)
     15 #sort distances and find indexes

---> 16         index=np.argsort(dist)
     17 #pick the kth closest indexes

     18         knearest=index[1:k]

NameError: global name 'numpy' is not defined
4

2 回答 2

1

由于您正在打电话np.some_function,因此您不必使用from numpy import *.

函数的局部定义必须与全局定义冲突,因此请尝试import numpy在函数内部移动:

def knn(Q, R, k, classvar):
    import numpy as np

如果它不起作用,则必须在某个地方使用numpy.some_function而不是np.some_function...

于 2013-11-02T14:04:44.297 回答
0

在导入包“numpy”之前

import numpy as np

您可以将 np 定义为全局

global np

代码将像这样

global np
import numpy as np

def foo():
    np.seterr(divide = 'ignore')

这对我有用,也可能对你有用!!!

于 2017-06-28T12:14:42.573 回答