0

我想将我的数据绘制为英寸的物理单位（也就是说，我有“点”中的数据，我可以将其转换为英寸）。（我查看了 matplotlib 转换教程。）在下面的代码中，我希望能够自动计算所需的 figsize。为了做到这一点，我硬编码了三个变量，WIDTH_SHIFT、HEIGHT_SHIFT 和 PORTION_OF_PLOT_INSIDE_THE_MARGINS。这些值适用于本示例，但是，随着数字变大，我需要增加 SHIFT 变量，并且可能还需要更改 PORTION 变量。如果可能的话，我想自动计算这些，或者找到一种替代方法。如果我将 WIDTH_SHIFT 和 HEIGHT_SHIFT 设置为零，则 ABCDEFG 和 F 节点框绘制不正确。它们落在轴外。部分绘制了 ABCDEFG 框。 

import matplotlib.pyplot as plt
from matplotlib.patches import BoxStyle, FancyBboxPatch


# Set-up the node data

node_width_ABCDEFG = 116/72  # inches
node_height_ABCDEFG = 28/72  # inches
node_center_x_ABCDEFG = 0 + 58/72  # inches
node_center_y_ABCDEFG = 3 + 14/72  # inches
lower_left_corner_ABCDEFG = [
    node_center_x_ABCDEFG - node_width_ABCDEFG/2,
    node_center_y_ABCDEFG - node_height_ABCDEFG/2]

node_width_F = 28/72  # inches
node_height_F = 28/72  # inches
node_center_x_F = 3 + 14/72  # inches
node_center_y_F = 0 + 14/72  # inches
lower_left_corner_F = [node_center_x_F - node_width_F/2,
                       node_center_y_F - node_height_F/2]

node_width_E = 28/72  # inches
node_height_E = 28/72  # inches
node_center_x_E = 5 + 14/72  # inches
node_center_y_E = 4 + 14/72  # inches
lower_left_corner_E = [node_center_x_E - node_width_E/2,
                       node_center_y_E - node_height_E/2]

node_width_C = 28/72  # inches
node_height_C = 28/72  # inches
node_center_x_C = 2.8 + 14/72  # inches  
node_center_y_C = 4.4 + 14/72  # inches
lower_left_corner_C = [node_center_x_C - node_width_C/2,
                       node_center_y_C - node_height_C/2]

# Compute size parameters needed to create the figure

# left edge of box ABCDEFG to right edge of box E
network_width_inches = (
    (node_center_x_E + node_width_E/2) - 
    (node_center_x_ABCDEFG - node_width_ABCDEFG/2))

# top edge of box C to bottom edge of box F
network_height_inches = (
    (node_center_y_C + node_height_C/2) - 
    (node_center_y_F - node_height_F/2))

print("network width inches: ", network_width_inches)
print("network height inches: ", network_height_inches)

WIDTH_SHIFT = .75  # width of left margin 
HEIGHT_SHIFT = 0.5  # height of bottom margin

# when these are zero, 
# the node boxes for ABCDEFG and F fall outside the axes
# WIDTH_SHIFT = 0  
# HEIGHT_SHIFT = 0

PORTION_OF_PLOT_INSIDE_THE_MARGINS = .7

fig_width_inches = 
    network_width_inches/PORTION_OF_PLOT_INSIDE_THE_MARGINS
fig_height_inches = 
    network_height_inches/PORTION_OF_PLOT_INSIDE_THE_MARGINS

# These enlargements are not big enough
# fig_width_inches = network_width_inches + 2*WIDTH_SHIFT
# fig_height_inches = network_height_inches + 2*HEIGHT_SHIFT

print("fig width inches: ", fig_width_inches)
print("fig height inches: ", fig_height_inches)

# Create the figure

fig = plt.figure(figsize=(fig_width_inches, fig_height_inches), dpi=72)

ax = plt.gca()

# Plotting in inches

# Create the node boxes

node_patch_ABCDEFG = FancyBboxPatch(
        [lower_left_corner_ABCDEFG[0] + WIDTH_SHIFT, 
         lower_left_corner_ABCDEFG[1] + HEIGHT_SHIFT],
        node_width_ABCDEFG,
        node_height_ABCDEFG,
        edgecolor="blue",
        facecolor="lightblue",
        linewidth=1,
        boxstyle=BoxStyle("round", 
                          pad=0,
                          rounding_size=.1),
        transform=fig.dpi_scale_trans
        )

node_patch_C = FancyBboxPatch(
        [lower_left_corner_C[0] + WIDTH_SHIFT, 
         lower_left_corner_C[1] + HEIGHT_SHIFT],
        node_width_C,
        node_height_C,
        edgecolor="blue",
        facecolor="lightblue",
        linewidth=1,
        boxstyle=BoxStyle("round", 
                          pad=0,
                          rounding_size=.1),
        transform=fig.dpi_scale_trans
        )

node_patch_E = FancyBboxPatch(
        [lower_left_corner_E[0] + WIDTH_SHIFT, 
         lower_left_corner_E[1] + HEIGHT_SHIFT],
        node_width_E,
        node_height_E,
        edgecolor="blue",
        facecolor="lightblue",
        linewidth=1,
        boxstyle=BoxStyle("round", 
                          pad=0,
                          rounding_size=.1),
        transform=fig.dpi_scale_trans
        )            

node_patch_F = FancyBboxPatch(
        [lower_left_corner_F[0] + WIDTH_SHIFT, 
         lower_left_corner_F[1] + HEIGHT_SHIFT],
        node_width_F,
        node_height_F,
        edgecolor="blue",
        facecolor="lightblue",
        linewidth=1,
        boxstyle=BoxStyle("round", 
                          pad=0,
                          rounding_size=.1),
        transform=fig.dpi_scale_trans
        )

ax.add_patch(node_patch_ABCDEFG)
ax.add_patch(node_patch_C)
ax.add_patch(node_patch_E)
ax.add_patch(node_patch_F)

# Add text labels for the nodes

plt.text(
    node_center_x_ABCDEFG + WIDTH_SHIFT,
    node_center_y_ABCDEFG + HEIGHT_SHIFT,
    "ABCDEFG",
    fontsize=12,
    color="black",
    fontname="Verdana",
    fontstyle="normal",
    horizontalalignment="center",
    verticalalignment="center",
    transform=fig.dpi_scale_trans
    )

plt.text(
    node_center_x_C + WIDTH_SHIFT,
    node_center_y_C + HEIGHT_SHIFT,
    "C",
    fontsize=12,
    color="black",
    fontname="Verdana",
    fontstyle="normal",
    horizontalalignment="center",
    verticalalignment="center",
    transform=fig.dpi_scale_trans
    )

plt.text(
    node_center_x_E + WIDTH_SHIFT,
    node_center_y_E + HEIGHT_SHIFT,
    "E",
    fontsize=12,
    color="black",
    fontname="Verdana",
    fontstyle="normal",
    horizontalalignment="center",
    verticalalignment="center",
    transform=fig.dpi_scale_trans
    )

plt.text(
    node_center_x_F + WIDTH_SHIFT,
    node_center_y_F + HEIGHT_SHIFT,
    "F",
    fontsize=12,
    color="black",
    fontname="Verdana",
    fontstyle="normal",
    horizontalalignment="center",
    verticalalignment="center",
    transform=fig.dpi_scale_trans
    )

plt.axis("equal")

#plt.axis("off")

plt.show()

fig.savefig("node_example.png", bbox_inches="tight")

正确绘制的节点框

在 subplots_adjust(0,0,1,1) 和 SHIFTS 到 0 和 PORTION 到 1 之后

从 savefig 中删除 bbox_inches="tight"

在 iPython 控制台中的 subplots_adjust(0,0,1,1) 之后

4

1 回答 1

0

感谢ImportanceOfBeingErnest，

解决方案是从图中删除边距，可以通过以下方式完成：

fig.subplots_adjust(0,0,1,1)

在这种情况下，不需要设置任何值来将节点图从边缘移到轴中或使图大于图本身。

为了在保存图窗时使用此功能，

bbox_inches

参数不应该用于

fig.savefig()

要使其与 iPython 控制台内联工作，请使用以下命令： 

%config InlineBackend.print_figure_kwargs = {'bbox_inches':None}
于 2019-06-21T03:04:53.733 回答