20

我想获取直方图 bin 中包含的数据列表。我正在使用 numpy 和 Matplotlib。我知道如何遍历数据并检查 bin 边缘。但是,我想为 2D 直方图执行此操作,并且执行此操作的代码相当难看。numpy 是否有任何结构可以使这更容易?

对于一维情况,我可以使用 searchsorted()。但是逻辑并没有那么好,我真的不想在不需要时对每个数据点进行二进制搜索。

大多数讨厌的逻辑是由于 bin 边界区域造成的。所有区域都有这样的边界:[左边缘,右边缘)。除了最后一个 bin,它有一个像这样的区域:[left edge, right edge]。

以下是一维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

以下是 2D 案例的一些示例代码:

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

有没有更清洁/更有效的方法来做到这一点?似乎 numpy 会为此提供一些东西。

4

2 回答 2

27

digitize,来自核心 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)     
于 2010-02-17T01:12:32.543 回答
6

怎么样:

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)]) 

出去:

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

用一些代码来处理边缘情况。

于 2010-02-16T22:08:35.513 回答