[英]Custom ticks for seaborn heatmap
I have some data that I would like to plot as a heatmap, it is essentially a 50x50 numpy array. 我有一些数据 ,我想绘制为热图,它本质上是一个50×50阵列numpy的。 As a result the heatmap axis labels range from 0 to 50, but actually I want the axis labels to go from -114 to 114 since this is the range of the data. 结果,热图轴标签的范围是0到50,但实际上我希望轴标签的范围是-114到114,因为这是数据的范围。 When I set the tick labels however, they end up being bunched up on the axes (see image). 但是,当我设置刻度线标签时,它们最终会被束缚在轴上(见图)。
When I put in the lines 当我排队时
ax.set_xticks(ticks)
ax.set_yticks(ticks)
The heatmap ends up getting scaled (see image). 热图最终会缩放(参见图片)。
I have put in my code and some sample data, maybe someone can spot what I have done wrong. 我已经输入了代码和一些示例数据,也许有人可以发现我做错了什么。
import sys
import numpy as np
import pandas as pd
import matplotlib
import matplotlib.pyplot as plt
import os
import cv2 as cv
import seaborn as sns;
filepath = sys.argv[1]
drive, path_and_file = os.path.splitdrive(filepath)
path, file = os.path.split(path_and_file)
line_width = 3
font = {'family' : 'sans',
'weight' : 'normal',
'size' : 18}
matplotlib.rc('font', **font)
bagnames = ["hex_events_only.bag"]
groundtruth = [-92, 0]
noise_levels = ["-1.000000"]
rewards = ["sos"]
gt_angle = np.arctan2(groundtruth[0], groundtruth[1])
gt_mag = np.linalg.norm(groundtruth, axis=0)
print(gt_angle, gt_mag)
for bagname in bagnames:
print "==========", bagname, "=========="
for reward in rewards:
print " ---", reward, "--- "
for noise_level in noise_levels:
filename = filepath + "data_field_" + bagname + "_" + reward + "_" + noise_level
print filename
n_samples = (pd.read_csv(filename, delimiter="\t", skiprows=1, names=["vx", "vy", "measure"])).values
x = n_samples[:, 0]
y = n_samples[:, 1]
z = n_samples[:, 2]
yrange = int(np.ptp(x))
xrange = int(np.ptp(y))
x_values = np.unique(x).size
y_values = np.unique(y).size
num_ticks = 10
ticks = np.linspace(int(-yrange/2.), int(yrange/2.), num_ticks, dtype=np.int)
img = np.reshape(z, (x_values, y_values))
img = img.T
img = cv.resize(img, (yrange, xrange))
savename = filepath + "hmap_" + bagname + "_" + reward + "_" + noise_level
fig, ax = plt.subplots()
img = cv.GaussianBlur(img, (5, 5), 0)
ax = sns.heatmap(img, cmap='viridis', yticklabels=ticks, xticklabels=ticks)
# ax.set_xticks(ticks)
# ax.set_yticks(ticks)
# ax.axvline(groundtruth[0], linestyle='--', c='r', linewidth=line_width)
# ax.axhline(groundtruth[1], linestyle='--', c='r', linewidth=line_width)
plt.show()
fig.savefig(savename + ".png", transparent=True, bbox_inches='tight', pad_inches=0)
plt.close()
https://1drv.ms/u/s!Ap0up1KFhZOughZ3dx9rwq-9yiF9 https://1drv.ms/u/s!Ap0up1KFhZOughZ3dx9rwq-9yiF9
@ImportanceOfBeingErnest pointed out to me that the approach of using Seaborn was wrong in the first place (see comments). @ImportanceOfBeingErnest向我指出,使用Seaborn的方法首先是错误的(请参阅评论)。 So I changed the approach, which now works exactly as I want it to. 因此,我更改了方法,该方法现在完全可以按我的意愿工作。 In case anyone else runs into this problem, the following code with generate a heatmap from data: 万一其他人遇到此问题,以下代码可从数据生成热图:
import numpy as np
import pandas as pd
import matplotlib
import matplotlib.pyplot as plt
import cv2 as cv
font = {'family' : 'sans',
'weight' : 'normal',
'size' : 18}
matplotlib.rc('font', **font)
filepath = "path/to/data/"
dataname = "data.txt"
filename = filepath + dataname
n_samples = (pd.read_csv(filename, delimiter="\t", skiprows=1, names=["x", "y", "value"])).values
x = n_samples[:, 0]
y = n_samples[:, 1]
z = n_samples[:, 2]
line_width = 2
yrange = int(np.ptp(x))
xrange = int(np.ptp(y))
x_values = np.unique(x).size
y_values = np.unique(y).size
num_ticks = 10
ticks = np.linspace(int(-yrange/2.), int(yrange/2.), num_ticks, dtype=np.int)
img = np.reshape(z, (x_values, y_values))
img = img.T
img = cv.resize(img, (yrange, xrange))
fig, ax = plt.subplots()
im = ax.imshow(img, cmap='viridis', extent=[-xrange/2., xrange/2., -yrange/2., yrange/2.])
ax.axvline(groundtruth[0], linestyle='--', c='r', linewidth=line_width)
ax.axhline(groundtruth[1], linestyle='--', c='r', linewidth=line_width)
ax.set_xlabel("$v_x$")
ax.set_ylabel("$v_y$")
cbar = fig.colorbar(im)
cbar.ax.set_yticklabels([''])
cbar.ax.set_ylabel('Reward')
fig.tight_layout()
savename = filepath + "hmap_" + bagname + "_" + reward + "_" + noise_level
fig.savefig(savename + ".pdf", transparent=True, bbox_inches='tight', pad_inches=0)
plt.close()
# plt.show()
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.