2013-09-29 238 views
20

我試圖停止註釋文本在我的圖中重疊。在接受的Matplotlib overlapping annotations答案中建議的方法看起來非常有前途,但是對於條形圖而言。我無法將「軸」方法轉換爲我想要執行的操作,但我不明白文本是如何排列的。Matplotlib重疊註釋/文本

import sys 
import matplotlib.pyplot as plt 


# start new plot 
plt.clf() 
plt.xlabel("Proportional Euclidean Distance") 
plt.ylabel("Percentage Timewindows Attended") 
plt.title("Test plot") 

together = [(0, 1.0, 0.4), (25, 1.0127692669427917, 0.41), (50, 1.016404709797609, 0.41), (75, 1.1043426359673716, 0.42), (100, 1.1610446924342996, 0.44), (125, 1.1685687930691457, 0.43), (150, 1.3486407784550272, 0.45), (250, 1.4013999168008104, 0.45)] 
together.sort() 

for x,y,z in together: 
    plt.annotate(str(x), xy=(y, z), size=8) 

eucs = [y for (x,y,z) in together] 
covers = [z for (x,y,z) in together] 

p1 = plt.plot(eucs,covers,color="black", alpha=0.5) 

plt.savefig("test.png") 

圖片(如果這個工程)可以發現here(此代碼):

image1

here(複雜):

image2

+0

另見http://stackoverflow.com/questions/14938541/how-to-improve-the-label-placement-for-matplotlib-scatter-chart-code-算法/ 15859652#15859652 – tacaswell

回答

40

我只是想在這裏發表另一種解決方案,一個小型圖書館,我寫來實現這樣的事情:https://github.com/Phlya/adjustText 過程的一個例子可以看這裏: enter image description here

這裏是示例圖像:

import matplotlib.pyplot as plt 
from adjustText import adjust_text 
import numpy as np 
together = [(0, 1.0, 0.4), (25, 1.0127692669427917, 0.41), (50, 1.016404709797609, 0.41), (75, 1.1043426359673716, 0.42), (100, 1.1610446924342996, 0.44), (125, 1.1685687930691457, 0.43), (150, 1.3486407784550272, 0.45), (250, 1.4013999168008104, 0.45)] 
together.sort() 

text = [x for (x,y,z) in together] 
eucs = [y for (x,y,z) in together] 
covers = [z for (x,y,z) in together] 

p1 = plt.plot(eucs,covers,color="black", alpha=0.5) 
texts = [] 
for x, y, s in zip(eucs, covers, text): 
    texts.append(plt.text(x, y, s)) 

plt.xlabel("Proportional Euclidean Distance") 
plt.ylabel("Percentage Timewindows Attended") 
plt.title("Test plot") 
adjust_text(texts, only_move='y', arrowprops=dict(arrowstyle="->", color='r', lw=0.5)) 
plt.show() 

enter image description here

如果你想有一個完美的身材,你可以反覆折騰一點。首先,讓我們使文本排斥線 - 爲此,我們使用scipy.interpolate.interp1d沿着它們創建大量虛擬點。

我們希望避免將標籤沿着x軸移動,因爲,爲什麼不爲了說明的目的去做。爲此,我們使用參數only_move={'points':'y', 'text':'y'}。如果我們只想在它們與文本重疊的情況下沿x軸移動它們,請使用move_only={'points':'y', 'text':'xy'}。在開始時,函數還會選擇文本相對於其原始點的最佳對齊方式,因此我們只希望它也沿着y軸發生,因此autoalign='y'。我們還減少了來自點的排斥力,以避免由於我們人爲避免線條而導致文本飛得太遠。總之:

from scipy import interpolate 
p1 = plt.plot(eucs,covers,color="black", alpha=0.5) 
texts = [] 
for x, y, s in zip(eucs, covers, text): 
    texts.append(plt.text(x, y, s)) 

f = interpolate.interp1d(eucs, covers) 
x = np.arange(min(eucs), max(eucs), 0.0005) 
y = f(x)  

plt.xlabel("Proportional Euclidean Distance") 
plt.ylabel("Percentage Timewindows Attended") 
plt.title("Test plot") 
adjust_text(texts, x=x, y=y, autoalign='y', 
      only_move={'points':'y', 'text':'y'}, force_points=0.15, 
      arrowprops=dict(arrowstyle="->", color='r', lw=0.5)) 
plt.show() 

enter image description here

+0

不錯的作品Phlya!你也可以添加這個答案或類似的東西https://stackoverflow.com/questions/9074996/matplotlib-how-to-annotate-point-on-a-scatter-automatically-placed-arrow – naught101

+0

謝謝,我'很高興你喜歡它!還有其他一些關於SO的問題是相關的,但我還沒有看到那個......我會盡力找時間爲它寫一個答案,但如果你願意的話,你可以自由地這麼做! – Phlya

+0

這真是太棒了,遊戲改變者 – dataflow

3

隨着大量我弄明白了。原來的解決方案再次獲得Matplotlib overlapping annotations的答案。

我不知道如何找到文本的確切寬度和高度。如果有人知道,請發佈改進(或者在方法中添加評論)。

import sys 
import matplotlib 
import matplotlib.pyplot as plt 
import numpy as np 

def get_text_positions(text, x_data, y_data, txt_width, txt_height): 
    a = zip(y_data, x_data) 
    text_positions = list(y_data) 
    for index, (y, x) in enumerate(a): 
     local_text_positions = [i for i in a if i[0] > (y - txt_height) 
          and (abs(i[1] - x) < txt_width * 2) and i != (y,x)] 
     if local_text_positions: 
      sorted_ltp = sorted(local_text_positions) 
      if abs(sorted_ltp[0][0] - y) < txt_height: #True == collision 
       differ = np.diff(sorted_ltp, axis=0) 
       a[index] = (sorted_ltp[-1][0] + txt_height, a[index][1]) 
       text_positions[index] = sorted_ltp[-1][0] + txt_height*1.01 
       for k, (j, m) in enumerate(differ): 
        #j is the vertical distance between words 
        if j > txt_height * 2: #if True then room to fit a word in 
         a[index] = (sorted_ltp[k][0] + txt_height, a[index][1]) 
         text_positions[index] = sorted_ltp[k][0] + txt_height 
         break 
    return text_positions 

def text_plotter(text, x_data, y_data, text_positions, txt_width,txt_height): 
    for z,x,y,t in zip(text, x_data, y_data, text_positions): 
     plt.annotate(str(z), xy=(x-txt_width/2, t), size=12) 
     if y != t: 
      plt.arrow(x, t,0,y-t, color='red',alpha=0.3, width=txt_width*0.1, 
       head_width=txt_width, head_length=txt_height*0.5, 
       zorder=0,length_includes_head=True) 

# start new plot 
plt.clf() 
plt.xlabel("Proportional Euclidean Distance") 
plt.ylabel("Percentage Timewindows Attended") 
plt.title("Test plot") 

together = [(0, 1.0, 0.4), (25, 1.0127692669427917, 0.41), (50, 1.016404709797609, 0.41), (75, 1.1043426359673716, 0.42), (100, 1.1610446924342996, 0.44), (125, 1.1685687930691457, 0.43), (150, 1.3486407784550272, 0.45), (250, 1.4013999168008104, 0.45)] 
together.sort() 

text = [x for (x,y,z) in together] 
eucs = [y for (x,y,z) in together] 
covers = [z for (x,y,z) in together] 

p1 = plt.plot(eucs,covers,color="black", alpha=0.5) 

txt_height = 0.0037*(plt.ylim()[1] - plt.ylim()[0]) 
txt_width = 0.018*(plt.xlim()[1] - plt.xlim()[0]) 

text_positions = get_text_positions(text, eucs, covers, txt_width, txt_height) 

text_plotter(text, eucs, covers, text_positions, txt_width, txt_height) 

plt.savefig("test.png") 
plt.show() 

創建http://i.stack.imgur.com/xiTeU.png enter image description here

更復雜的圖形現在http://i.stack.imgur.com/KJeYW.png,仍然是一個有點玄乎,但好多了! enter image description here

+0

請驗證我沒有搞清楚編輯。 – tacaswell

+0

,並記得在接受你的答案時,它會讓你。 – tacaswell

+0

和'get_window_extent()'是您想要 – tacaswell