使用两个不同颜色的数据集创建matplotlib热图

时间:2016-03-01 15:51:47

标签: python matplotlib heatmap

我目前有两个大型数据集,我想比较它们。我有它们分开,一个是红色,一个是蓝色,但我想并排显示红色和蓝色。我怎么能这样做?

我目前的代码是:

column_labels = list(heatmap_ylabels)
row_labels = list(heatmap_xlabels)
fig, ax = plt.subplots()
heatmap = ax.pcolor(data, cmap=plt.cm.Reds)

ax.set_xticks(np.arange(9+0.5))
ax.set_yticks(np.arange(140+0.5))

ax.invert_yaxis()
ax.xaxis.tick_top()
ax.set_xticklabels(row_labels, minor=False)
ax.set_yticklabels(column_labels, minor=False)
#plt.show()
plt.savefig('n1_heatmap')
plt.clf()

column_labels = list(heatmap_ylabels)
row_labels = list(heatmap_xlabels)
fig, ax = plt.subplots()
heatmap = ax.pcolor(data1, cmap=plt.cm.Blues)

ax.set_xticks(np.arange(9+0.5))
ax.set_yticks(np.arange(140+0.5))

ax.invert_yaxis()
ax.xaxis.tick_top()
ax.set_xticklabels(row_labels, minor=False)
ax.set_yticklabels(column_labels, minor=False)
plt.savefig('n2_heatmap')
plt.clf()

datadata1都由140个不同的列表组成,其中的信息是从280个不同的文件中提取的,有没有办法我仍然可以使用这两个列表来创建一个热图,它将显示这些数据在同一图中?

因此,例如我的热图将是/红色/蓝色/红色/蓝色等

以下是我的热图的示例:

enter image description here

编辑:

虽然没有准确显示我想要的内容,但我已经制作了前两个热图之间差异值的热图。

例如:y2 = np.subtract(y, y1)

data2.append(y2)
column_labels = list(heatmap_ylabels)
row_labels = list(heatmap_xlabels)
fig, ax = plt.subplots()
heatmap = ax.pcolor(data2, cmap=plt.cm.bwr)

ax.set_xticks(np.arange(9+0.5))
ax.set_yticks(np.arange(140+0.5))

ax.invert_yaxis()
ax.xaxis.tick_top()
ax.set_xticklabels(row_labels, minor=False)
ax.set_yticklabels(column_labels, minor=False)
plt.savefig('diff_heatmap')
plt.clf()

1 个答案:

答案 0 :(得分:3)

正如@jeanrjc所提到的,这在概念上与a previously-asked question非常相似。但是,在您的案例中如何应用该方法可能并不明显。

这是一个最小的例子,用两个不同的彩色图标绘两个具有“并排”相同形状的阵列。关键是要独立绘制两个蒙版数组。为了创建这些掩码数组,我们将使用双列数来创建新数组,并且每隔一列掩盖一次。

这是一个简单的例子(请注意,有几种方法可以创建蒙版数组模式):

import numpy as np
import matplotlib.pyplot as plt

# Generate data
nrows, ncols = 20, 5
x = np.random.random((nrows, ncols))
y = np.random.random((nrows, ncols))

# Make data for display
mask = np.array(nrows * [ncols * [False, True]], dtype=bool)
red = np.ma.masked_where(mask, np.repeat(x, 2, axis=1))

mask = np.array(nrows * [ncols * [True, False]], dtype=bool)
blue = np.ma.masked_where(mask, np.repeat(y, 2, axis=1))

# Make a side-by-side plot
fig, ax = plt.subplots()
ax.pcolormesh(red, cmap='Reds')
ax.pcolormesh(blue, cmap='Blues')
plt.show()

enter image description here

如果我们想制作一个更好的版本,我们可以做类似的事情:

import numpy as np
import matplotlib.pyplot as plt

# Generate data
nrows, ncols = 20, 5
x = np.exp(np.random.normal(0, 0.8, (nrows, ncols)))
y = np.exp(np.random.normal(0, 1, (nrows, ncols)))

# Make data for display
mask = np.array(nrows * [ncols * [False, True]], dtype=bool)
red = np.ma.masked_where(mask, np.repeat(x, 2, axis=1))

mask = np.array(nrows * [ncols * [True, False]], dtype=bool)
blue = np.ma.masked_where(mask, np.repeat(y, 2, axis=1))

# Make a side-by-side plot
fig, ax = plt.subplots()
redmesh = ax.pcolormesh(red, cmap='Reds')
bluemesh = ax.pcolormesh(blue, cmap='Blues')

# Make things a touch fancier
ax.set(xticks=np.arange(1, 2 * ncols, 2),
       yticks=np.arange(nrows) + 0.5,
       xticklabels=['Column ' + letter for letter in 'ABCDE'],
       yticklabels=['Row {}'.format(i+1) for i in range(nrows)])

ax.set_title('Side-by-Side Plot', y=1.07)
ax.xaxis.tick_top()
ax.yaxis.tick_left()
ax.tick_params(direction='out')

# Add dual colorbars
fig.subplots_adjust(bottom=0.05, right=0.78, top=0.88)
cbar = fig.colorbar(redmesh, cax=fig.add_axes([0.81, 0.05, 0.04, 0.83]))
cbar.ax.text(0.55, 0.1, 'Variable 1', rotation=90, ha='center', va='center',
             transform=cbar.ax.transAxes, color='gray')
cbar = fig.colorbar(bluemesh, cax=fig.add_axes([0.9, 0.05, 0.04, 0.83]))
cbar.ax.text(0.55, 0.1, 'Variable 2', rotation=90, ha='center', va='center',
             transform=cbar.ax.transAxes, color='gray')

# Make the grouping clearer
ax.set_xticks(np.arange(0, 2 * ncols, 2), minor=True)
ax.grid(axis='x', ls='-', color='gray', which='minor')
ax.grid(axis='y', ls=':', color='gray')

plt.show()

enter image description here