使用滑块时如何更新直方图？

Question

我想为正态分布构建直方图，并在均值、标准差和样本量发生变化时更新绘图；类似于帖子here。

但是，我在 update 函数上遇到了困难。在上面的例子中

l, = plot(f(S, 1.0, 1.0))

和

def update(val):
    l.set_ydata(f(S, sGmax.val, sKm.val))

已使用，但在绘制直方图时必须如何更改？所以，我不确定如何使用来自plt.hist 的返回值，将它们正确传递给update，然后相应地更新绘图。谁能解释一下？

这是我的代码：

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.widgets import Slider


def update(val):
    mv = smean.val
    stdv = sstd.val
    n_sample = round(sn.val)
    # what needs to go here? how to replace xxx
    xxx(np.random.normal(mv, stdv, n_sample))
    plt.draw()


ax = plt.subplot(111)
plt.subplots_adjust(left=0.25, bottom=0.25)

m0 = -2.5
std0 = 1
n0 = 1000
n_bins0 = 20

nd = np.random.normal(m0, std0, n0)

# what needs to be returned here?
plt.hist(nd, normed=True, bins=n_bins0, alpha=0.5)

axcolor = 'lightgray'
axmean = plt.axes([0.25, 0.01, 0.65, 0.03], axisbg=axcolor)
axstd = plt.axes([0.25, 0.06, 0.65, 0.03], axisbg=axcolor)
axssize = plt.axes([0.25, 0.11, 0.65, 0.03], axisbg=axcolor)

smean = Slider(axmean, 'Mean', -5, 5, valinit=m0)
sstd = Slider(axstd, 'Std', 0.1, 10.0, valinit=std0)
sn = Slider(axssize, 'n_sample', 10, 10000, valinit=n0)

smean.on_changed(update)
sstd.on_changed(update)
sn.on_changed(update)

plt.show()

Answer 1

一种选择是清除轴并重新绘制直方图。另一个选项，更符合 matplotlib 滑块 example 的 l.set_value 方法的精神，将使用 numpy 生成直方图数据，使用条形图并使用 bar.set_height 和 bar.set_x 使用 rescale 更新它轴上。完整的例子是：

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.widgets import Slider


def update(val):
    mv = smean.val
    stdv = sstd.val
    n_sample = round(sn.val)
    nd = np.random.normal(loc=mv, scale=stdv, size=n_sample)
    #Update barchart height and x values
    hist, bins = np.histogram(nd, normed=True, bins=n_bins0)
    [bar.set_height(hist[i]) for i, bar in enumerate(b)]
    [bar.set_x(bins[i]) for i, bar in enumerate(b)]
    ax.relim()
    ax.autoscale_view()
    plt.draw()


def reset(event):
    mv.reset()
    stdv.reset()
    n_sample.reset()


ax = plt.subplot(111)
plt.subplots_adjust(left=0.25, bottom=0.25)

m0 = -2.5
std0 = 1
n0 = 1000
n_bins0 = 20

nd = np.random.normal(m0, std0, n0)
hist, bins = np.histogram(nd, normed=True, bins=n_bins0)
b = plt.bar(bins[:-1], hist, width=.3)

axcolor = 'lightgray'
axmean = plt.axes([0.25, 0.01, 0.65, 0.03], axisbg=axcolor)
axstd = plt.axes([0.25, 0.06, 0.65, 0.03], axisbg=axcolor)
axssize = plt.axes([0.25, 0.11, 0.65, 0.03], axisbg=axcolor)

smean = Slider(axmean, 'Mean', -5, 5, valinit=m0)
sstd = Slider(axstd, 'Std', 0.1, 10.0, valinit=std0)
sn = Slider(axssize, 'n_sample', 10, 10000, valinit=n0)

smean.on_changed(update)
sstd.on_changed(update)
sn.on_changed(update)

plt.show()

更新：

版本使用清轴（ax.cla()）并重绘ax.hist(...)，

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.widgets import Slider


def update(val):
    mv = smean.val
    stdv = sstd.val
    n_sample = round(sn.val)
    nd = np.random.normal(loc=mv, scale=stdv, size=n_sample)
    #Redraw histogram
    ax.cla()
    ax.hist(nd, normed=True, bins=n_bins0, alpha=0.5)
    plt.draw()


def reset(event):
    mv.reset()
    stdv.reset()
    n_sample.reset()


ax = plt.subplot(111)
plt.subplots_adjust(left=0.25, bottom=0.25)

m0 = -2.5
std0 = 1
n0 = 1000
n_bins0 = 20

nd = np.random.normal(m0, std0, n0)
plt.hist(nd, normed=True, bins=n_bins0, alpha=0.5)

axcolor = 'lightgray'
axmean = plt.axes([0.25, 0.01, 0.65, 0.03], axisbg=axcolor)
axstd = plt.axes([0.25, 0.06, 0.65, 0.03], axisbg=axcolor)
axssize = plt.axes([0.25, 0.11, 0.65, 0.03], axisbg=axcolor)

smean = Slider(axmean, 'Mean', -5, 5, valinit=m0)
sstd = Slider(axstd, 'Std', 0.1, 10.0, valinit=std0)
sn = Slider(axssize, 'n_sample', 10, 10000, valinit=n0)

smean.on_changed(update)
sstd.on_changed(update)
sn.on_changed(update)

plt.show()

Answer 2

您可以在 matplotlib.container.BarContainer 对象上使用 remove() 函数

fig, ax = plt.subplots(constrained_layout=True, figsize=(9, 3.5))

_,_,q = ax.hist([1,2,3,4,4,4,3])
fig.canvas.draw()

q.remove()
_,_,q = ax.hist([1,2,3,4,1,1,3])
fig.canvas.draw()