如何在直方图箱中获取数据

Question

我想得到一个包含在直方图箱中的数据列表。我用的是numpy和Matplotlib。我知道如何遍历数据和检查垃圾箱边缘。不过，我想这样做的二维直方图和代码这样做是相当难看。numpy有什么结构可以让这更简单吗？

对于1D情况，我可以使用searchsorted（）。但是逻辑并没有那么好，我也不想在不需要的时候对每个数据点进行二进制搜索。

大多数令人讨厌的逻辑是由于垃圾箱边界区域。所有区域都有这样的边界：[左边缘，右边缘]。除了最后一个箱子，它有一个这样的区域：[左边缘，右边缘]。

下面是1D案例的一些示例代码：

import numpy as np

data = [0, 0.5, 1.5, 1.5, 1.5, 2.5, 2.5, 2.5, 3]

hist, edges = np.histogram(data, bins=3)

print 'data =', data
print 'histogram =', hist
print 'edges =', edges

getbin = 2  #0, 1, or 2

print '---'
print 'alg 1:'

#for i in range(len(data)):
for d in data:
    if d >= edges[getbin]:
        if (getbin == len(edges)-2) or d < edges[getbin+1]:
            print 'found:', d
        #end if
    #end if
#end for

print '---'
print 'alg 2:'

for d in data:
    val = np.searchsorted(edges, d, side='right')-1
    if val == getbin or val == len(edges)-1:
        print 'found:', d
    #end if
#end for

下面是二维情况的一些示例代码：

import numpy as np

xdata = [0, 1.5, 1.5, 2.5, 2.5, 2.5, \
         0.5, 0.5, 0.5, 0.5, 1.5, 1.5, 1.5, 1.5, 1.5, 2.5, 2.5, 2.5, 2.5, 2.5, 2.5, \
         0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 1.5, 1.5, 1.5, 1.5, 1.5, 1.5, 1.5, 1.5, 2.5, 2.5, 2.5, 2.5, 2.5, 2.5, 2.5, 2.5, 3]
ydata = [0, 5,5, 5, 5, 5, \
         15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, 15, \
         25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 25, 30]

xbins = 3
ybins = 3
hist2d, xedges, yedges = np.histogram2d(xdata, ydata, bins=(xbins, ybins))

print 'data2d =', zip(xdata, ydata)
print 'hist2d ='
print hist2d
print 'xedges =', xedges
print 'yedges =', yedges

getbin2d = 5  #0 through 8

print 'find data in bin #', getbin2d

xedge_i = getbin2d % xbins
yedge_i = int(getbin2d / xbins) #IMPORTANT: this is xbins

for x, y in zip(xdata, ydata):
    # x and y left edges
    if x >= xedges[xedge_i] and y >= yedges[yedge_i]:
        #x right edge
        if xedge_i == xbins-1 or x < xedges[xedge_i + 1]:
            #y right edge
            if yedge_i == ybins-1 or y < yedges[yedge_i + 1]:
                print 'found:', x, y
            #end if
        #end if
    #end if
#end for

有没有更干净/更有效的方法来做这件事？看来努比会有办法的。

Answer 1

比如说：

data = numpy.array([0, 0.5, 1.5, 1.5, 1.5, 2.5, 2.5, 2.5, 3])

hist, edges = numpy.histogram(data, bins=3)

for l, r in zip(edges[:-1], edges[1:]):
   print(data[(data > l) & (data < r)]) 

输出：

[ 0.5]
[ 1.5  1.5  1.5]
[ 2.5  2.5  2.5]

用一点代码来处理边缘情况。

Answer 2

^{}，来自core NumPy，将为您提供直方图中每个值所属的bin的索引：

import numpy as NP
A = NP.random.randint(0, 10, 100)

bins = NP.array([0., 20., 40., 60., 80., 100.])

# d is an index array holding the bin id for each point in A
d = NP.digitize(A, bins)