Python绘制直方图时如何将区间外数据归入最外侧bin？

核心实现逻辑是提前对输入数据做截断处理，无需修改原有bin设置、绘图逻辑，就能把超出范围的数据分别归入最左、最右侧bin。

具体实现步骤 #

1. 先计算你预设的直方图区间边界
2. 对原始数据做截断：所有小于左边界的数值替换为左边界，所有大于等于右边界的数值替换为略小于右边界的值（适配直方图左闭右开的统计规则）
3. 用处理后的数据生成直方图即可

可直接复用的修改后代码 #

#sort out N by histogram
import numpy as np # 若已导入可删除该行
binwidth = 2.5*sig_Q
low_center = min(Q_exp) + binwidth
def set_range(first_bin = low_center, bin_width = binwidth, Nbins = 10):
     """
     helper function to set the range and bin width
     input : first_bin =  bin_center of the first bin, bin_width = the bin width, Nbins = total number of bins
     returns: a tuple that you can use in the range key word when defining a histogram.
     NOTE: for the histogram use the same number of bins:

 example:  h = histo( r, range = set_range(-5., 1, 11), bins = 11)

 this created a histogram where the first bin is centered at -5. , the next at -4. etc. a total of 11 bins are
 created and the bin center of the last one is at 5. = first_bin + (Nbins-1)*bin_width
     """
 rmin = first_bin - bin_width/2.
 rmax = rmin + Nbins*bin_width
 return (rmin,rmax)

# -------------------新增修改部分开始-------------------
hist_range = set_range(low_center, bin_width=binwidth, Nbins=10)
rmin, rmax = hist_range
# 截断数据，极小偏移量避免右边界点被丢弃，数值远小于bin宽不影响统计结果
Q_exp_clipped = np.clip(Q_exp, rmin, rmax - 1e-12)
# 无numpy环境可替换为以下代码：
# Q_exp_clipped = [max(rmin, min(x, rmax - 1e-12)) for x in Q_exp]
# -------------------新增修改部分结束-------------------

h = B.histo(Q_exp_clipped, range = hist_range, bins = 10)
hx = h.bin_center
hy = h.bin_content
B.pl.ylabel("Counts", fontsize = 20)
B.pl.xlabel("Gaussian Deviates", fontsize = 20)
B.pl.title("Monte Carlo Millikan Oil-Drop Simulation", fontsize = 22)
h.plot()
B.pl.show()

补充说明 #

• 原有bin的中心位置、宽度、数量均未发生变化，不影响直方图的展示逻辑
• 1e-12的偏移量可根据你的数据精度调整，只要远小于bin宽度即可，避免对统计结果产生干扰

内容的提问来源于stack exchange，提问作者Valuska