Performance 对numpy数组子维的python操作
许多numpy函数提供了在轴=参数的特定轴上操作的选项。我的问题是Performance 对numpy数组子维的python操作,performance,pandas,numpy,multidimensional-array,Performance,Pandas,Numpy,Multidimensional Array,许多numpy函数提供了在轴=参数的特定轴上操作的选项。我的问题是 如何实现这种“沿轴”操作?或者,更直接的问题 我如何有效地编写自己的函数来提供类似的选项 我注意到numpy提供了一个函数,如果基本函数输入是1-D数组,该函数将用作答案 但是如果我的基函数需要多维输入呢?例如,沿前二维(5,6)求形状为(5,6,2,3,4)的np矩阵A的二维移动平均值B?就像一个普通函数B=f_移动_平均值(a,轴=(0,1)) 我当前的解决方案是使用numpy.swapaxes和numpy.reforme来
import pandas as pd
import numpy as np
def nanmoving_mean(data,window,axis=0):
kw = {'center':True,'window':window,'min_periods':1}
if len(data.shape)==1:
return pd.Series(data).rolling(**kw).mean().as_matrix()
elif len(data.shape)>=2:
tmp = np.swapaxes(data,0,axis)
tmpshp = tmp.shape
tmp = np.reshape( tmp, (tmpshp[0],-1), order='C' )
tmp = pd.DataFrame(tmp).rolling(**kw).mean().as_matrix()
tmp = np.reshape( tmp, tmpshp, order='C' )
return np.swapaxes(tmp,0,axis)
else:
print('Invalid dimension!')
return None
data = np.random.randint(10,size=(2,3,6))
print(data)
nanmoving_mean(data,window=3,axis=2)
这是问题2的一种常见/有效的实施方式吗?欢迎任何改进/建议/新方法
另外,我之所以在这里涉及pandas,是因为它的rolling(…).mean()方法能够正确处理nan数据
编辑:
我想问这个问题的另一种方式可能是:在“动态”维度数上循环的语法是什么?不要过多地讨论你的问题,这里是
沿轴应用功能的关键部分(通过Ipython查看)
它们构造了两个不同的索引对象,i
和ind
。假设我们指定轴=2,那么这个代码会
outarr[i,j,l] = func1d( arr[i,j,:,l], ...)
对于i
、j
和l
的所有可能值。所以有很多代码用于一个非常基本的迭代计算
ind = [0]*(nd-1) # ind is just a nd-1 list
i = zeros(nd, 'O') # i is a 1d array with a `slice` object
i[axis] = slice(None, None)
我不熟悉熊猫的滚动。但是有许多numpy
滚动平均问题scipy.signal.convolve2d
可能有用np.lib.stride\u技巧。as\u stride
也被使用
使用重塑
和交换轴
(或转置
)来降低维度空间的复杂性的想法也是很好的
(这不是一个解决方案;而是扔掉脑海中出现的一些想法,记住其他的“移动平均线”问题。现在进行更多的研究已经太迟了。)我们可以使用
基本步骤是:
- 作为预处理步骤,将
NaNs
替换为0s
,因为我们需要对输入数据进行加窗求和
- 获取数据值的加窗求和
还有
NaNs
的掩码。我们将使用边界元素作为零
- 从窗口大小中减去
NaNs
的窗口计数,以获得负责求和的有效元素的计数
- 对于边界元素,我们将逐渐使用较小的元素来计算总和
现在,这些区间求和
也可以通过相对更有效的方法获得。另一个好处是,我们可以指定执行这些求和/平均的轴
有了Scipy的2D卷积
和1D均匀滤波器
的混合,我们将有下面列出的几种方法
导入相关的Scipy函数-
from scipy.signal import convolve2d as conv2
from scipy.ndimage.filters import uniform_filter1d as uniff
方法#1:
def nanmoving_mean_numpy(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
value_sums = conv2(data1.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
nan_sums = conv2(nan_mask.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
value_sums.shape = data.shape
nan_sums.shape = data.shape
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW), b_sizes[::-1] ))
return value_sums/(count - nan_sums)
def nanmoving_mean_numpy_v2(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
value_sums = uniff(data1,size=W, axis=-1, mode='constant')*W
nan_sums = conv2(nan_mask.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
nan_sums.shape = data.shape
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW,dtype=int), b_sizes[::-1] ))
out = value_sums/(count - nan_sums)
out = np.where(np.isclose( count, nan_sums), np.nan, out)
return out
def nanmoving_mean_numpy_v3(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
nan_avgs = uniff(nan_mask.astype(float),size=W, axis=-1, mode='constant')
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW), b_sizes[::-1] ))
scale = ((count/float(W)) - nan_avgs)
out = uniff(data1,size=W, axis=-1, mode='constant')/scale
out = np.where(np.isclose( scale, 0), np.nan, out)
return out
方法#2:
def nanmoving_mean_numpy(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
value_sums = conv2(data1.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
nan_sums = conv2(nan_mask.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
value_sums.shape = data.shape
nan_sums.shape = data.shape
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW), b_sizes[::-1] ))
return value_sums/(count - nan_sums)
def nanmoving_mean_numpy_v2(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
value_sums = uniff(data1,size=W, axis=-1, mode='constant')*W
nan_sums = conv2(nan_mask.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
nan_sums.shape = data.shape
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW,dtype=int), b_sizes[::-1] ))
out = value_sums/(count - nan_sums)
out = np.where(np.isclose( count, nan_sums), np.nan, out)
return out
def nanmoving_mean_numpy_v3(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
nan_avgs = uniff(nan_mask.astype(float),size=W, axis=-1, mode='constant')
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW), b_sizes[::-1] ))
scale = ((count/float(W)) - nan_avgs)
out = uniff(data1,size=W, axis=-1, mode='constant')/scale
out = np.where(np.isclose( scale, 0), np.nan, out)
return out
方法#3:
def nanmoving_mean_numpy(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
value_sums = conv2(data1.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
nan_sums = conv2(nan_mask.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
value_sums.shape = data.shape
nan_sums.shape = data.shape
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW), b_sizes[::-1] ))
return value_sums/(count - nan_sums)
def nanmoving_mean_numpy_v2(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
value_sums = uniff(data1,size=W, axis=-1, mode='constant')*W
nan_sums = conv2(nan_mask.reshape(-1,N),np.ones((1,W)),'same', boundary='fill')
nan_sums.shape = data.shape
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW,dtype=int), b_sizes[::-1] ))
out = value_sums/(count - nan_sums)
out = np.where(np.isclose( count, nan_sums), np.nan, out)
return out
def nanmoving_mean_numpy_v3(data, W): # data: input array, W: Window size
N = data.shape[-1]
hW = (W-1)//2
nan_mask = np.isnan(data)
data1 = np.where(nan_mask,0,data)
nan_avgs = uniff(nan_mask.astype(float),size=W, axis=-1, mode='constant')
b_sizes = hW+1+np.arange(hW) # Boundary sizes
count = np.hstack(( b_sizes , W*np.ones(N-2*hW), b_sizes[::-1] ))
scale = ((count/float(W)) - nan_avgs)
out = uniff(data1,size=W, axis=-1, mode='constant')/scale
out = np.where(np.isclose( scale, 0), np.nan, out)
return out
运行时测试
数据集#1:
数据集#2[更大的数据集]:
In [811]: # Create random input array and insert NaNs
...: data = np.random.randint(10,size=(120,130,160)).astype(float)
...:
...: # Add 10% NaNs across the data randomly
...: idx = np.random.choice(data.size,size=int(data.size*0.1),replace=0)
...: data.ravel()[idx] = np.nan
...:
In [812]: %timeit nanmoving_mean(data,window=W,axis=2)
...: %timeit nanmoving_mean_numpy(data, W)
...: %timeit nanmoving_mean_numpy_v2(data, W)
...: %timeit nanmoving_mean_numpy_v3(data, W)
...:
1 loops, best of 3: 796 ms per loop
1 loops, best of 3: 486 ms per loop
1 loops, best of 3: 275 ms per loop
10 loops, best of 3: 161 ms per loop
我不知道在哪里看沿轴应用的代码,但它是如何构造I和ind的?@ShichuZhu如果你在寻找性能,apply\u沿轴应用
不会有帮助。@Divakar如果基函数只处理1D,那么在所有其他维度上循环是唯一的方法。除非修改基函数以包含向量化操作,否则我猜@ShichuZhu无需循环。有许多矢量化选项可用于在windows中求和元素。