Code interpretation of myplt box diagram
Group 47
Student(s)
- Yichen Li, liyichen18@lzu.edu.cn
- Yan Miao, miaoy2018@lzu.edu.cn
Myplt box diagram operation code explanation document and instruction
Introduction: the myplt class of boxplot can be roughly divided into three main code blocks, info_boxplot(), histobox_plot() and creative_ boxplot(). I will introduce it in four parts.
the package needed
#!/usr/bin/python3
# -*- coding: UTF-8 -*-
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np[TOC]
Now we begin
The explanation of cal_Quantile(),cal_max_min(),find_ex_point(),cul_all_max_min() and cul_step() in the class myplt
Source code of cal_Quantile
#!/usr/bin/python3
# -*- coding: UTF-8 -*-
def cal_Quantile(da):
if da.count()%2==0:
p_50=(da[int(da.count()/2-0.5)]+da[int(da.count()/2)])/2
da1=da[:int(da.count()/2)]
else:
p_50=da[int(da.count()/2)]
da1=da[:int((da.count()+1)/2)]
if da1.count()%2==0:
p_75=(da1[int(da1.count()/2-0.5)]+da1[int(da1.count()/2)])/2
da2=da[int(da.count()/2):]
else:
p_75=da1[int(da1.count()/2)]
da2=da[int((da.count()+1)/2):]
da2.reset_index(drop=True, inplace=True)
if da2.count()%2==0:
p_25=(da2[int(da2.count()/2-0.5)]+da2[int(da2.count()/2)])/2
else:
p_25=da2[int(da2.count()/2)]
da3=da[int(da1.count()/2):(int(da2.count()/2)+int(da.count()/2))]
da3.reset_index(drop=True, inplace=True)
return (p_25,p_50,p_75,da3)The input data is a single column series data da sorted by size.
The first if gets P_50 according to the median rule, that is, the median (odd number takes the ((da number + 1) / 2) and even number takes item (da number / 2)).
Then bisection da to get the first half da1.
Taking da1 as the total data set and taking the median rule again, so that p_75 is obtained.
After that it continues to bisect da to get the second half of da2.
According to the same rules, p_25 is obtained by taking da2 as the total data set.
da3 is the data set between p_25 and p_50, which will be used later in calculating 5% (cal_per5) to draw complex graphs.
The function returns three int data p_25, p_50 and p_75, the quartile, and a series data column da_3.
def cal_max_min(da,p_25,p_75):
ma=da[da<(p_25+(p_75-p_25)*1.5)].max()
mi=da[da>(p_25-(p_75-p_25)*1.5)].min()
return (ma,mi)The input data is a single column series data da sorted by size, and 25% and 75% of the quartile,p_25,p_75.
The returned values are the maximum and minimum values within the reasonable range of the boxplot.
It is used to draw the upper and lower edges of the boxplot and calculate the outliers.
def find_ex_point(data):
da=pd.Series(data).astype(int)
da=da.sort_values(ascending=False)
da.reset_index(drop=True, inplace=True)
p_25,p_50,p_75=myplt.cal_Quantile(da)[:3]
da_ex=da[(da<(p_25-(p_75-p_25)*1.5)) | (da>(p_25+(p_75-p_25)*1.5))]
return da_exThe input data is single-layer list data, and the previous cal_Quantile() function will be called in the function body.
After calculation, a single column series data is returned, and the content is the abnormal value in the data,da_ex.
def cul_all_max_min(data_l):
data_l=[[int(i) for i in k] for k in data_l]
return (max([max(i) for i in data_l]),min([min(i) for i in data_l]))The input data is multi-layer list data_ l.
The first step is to convert all the data into int data, and then calculate the maximum and minimum values of all the data.
Return two int data, that is, the maximum and minimum of all data, and they will be used to set the common Y-axis coordinates in creative_boxplot().
def cul_step(da_ma,da_mi):
n=(da_ma-da_mi)/4
count=0
while n>10:
count+=1
n/=10
step=int(n)*10**count
return stepThe input data are the maximum value da_ma and the minimum value da_mi in the current data column da.
The return value is the appropriate step size for the y-axis interval.
Used to set the y-axis coordinates in creative_boxplot().
def info_boxplot(ax,data_l,multiplebox=True,linecolor='black',pointcolor='black'):
ax.cla()
lenth=len(data_l)
ax.set_xlim((0,lenth+1))
tick_x=range(1,lenth+1)
ax.set_xticks(tick_x)
ax.set_xticklabels([str(i) for i in tick_x])
def cal_per5(da3):
per_5=[]
add=da3.count()*0.1
count=0
for i in range(9):
count+=add
per_5.append(da3[int(count-1)])
return per_5
def draw_boxplot(ax,mi,ma,p_25,p_50,p_75,multiplebox,da_ex,per_5,k,linecolor='black',pointcolor='black'):
ax.vlines(k,mi,p_25,color=linecolor,linewidth=1)
ax.vlines(k,p_75,ma,color=linecolor,linewidth=1)
ax.hlines(mi,k-0.15,k+0.15,color=linecolor,linewidth=1)
ax.hlines(ma,k-0.15,k+0.15,color=linecolor,linewidth=1)
rect = plt.Rectangle((k-0.25,p_25),0.5,p_75-p_25,linewidth=1, edgecolor=linecolor, facecolor='none')
ax.add_patch(rect)
#exception
for i in da_ex.values:
ax.text(k,i , "o",color=pointcolor, fontsize=5, verticalalignment="center",horizontalalignment='center')
#ax.scatter(k,i,color='white', marker='o', edgecolors='black')
ax.vlines(k,da_ex.min(),da_ex.max(),color='white',linewidth=1,zorder=0)
if (not multiplebox):
ax.hlines(p_50,k-0.25,k+0.25,color=linecolor,linewidth=1)
else:
#5%
for i in per_5:
ax.hlines(i,k-0.25,k+0.25,color=linecolor,linewidth=1)
ax.hlines(p_50,k-0.25,k+0.25,color=linecolor,linewidth=3)
def main_box(ax,data,multiplebox,k,linecolor,pointcolor):
da=pd.Series(data).astype(int)
da=da.sort_values(ascending=False)
da.reset_index(drop=True, inplace=True)
p_25,p_50,p_75,da3=myplt.cal_Quantile(da)
per_5=cal_per5(da3)
per_5[4]=p_50
ma,mi=myplt.cal_max_min(da,p_25,p_75)
da_ex=da[(da<(p_25-(p_75-p_25)*1.5)) | (da>(p_25+(p_75-p_25)*1.5))]
draw_boxplot(ax,mi,ma,p_25,p_50,p_75,multiplebox,da_ex,per_5,k,linecolor,pointcolor)
for k in range(lenth):
main_box(ax,data_l[k],multiplebox,k+1,linecolor,pointcolor)before cal_per5：Empty the canvas and set the x-axis ruler and label.
cal_per5：The input is the pandas interval da3, and the da3 interval is equally divided into ten parts according to the number of numbers. The list type per_5 records the nine dividing lines in the middle, which can also be understood as the y-axis scale that should be drawn. It is used to draw the 5% dividing line part of complex graph.
draw_boxplot：Input all kinds of parameters needed to draw the current box diagram, and the function is to draw a single box diagram. When multiplebox is true, nine more 5% boundaries are drawn to draw a complex box diagram.
main_box：The general manager to call various function functions. The input parameter (K+1) is the X-axis coordinate of the drawing line.
st=>start: Empty the canvas and set the X-axis
e=>end: Data input into draw_boxplot() for drawing
op1=>operation: The for loop makes every data entered into main_box() as a single-layer list data, and (k+1) is the x-axis scale
op2=>operation: The main_box() first converts list data into single column Series type data, and then calls corresponding functions after sorting
op3=>operation: Calculate the quartile int data p_25, p_50, p_75, series data da3,
5% list data per_5,
The maximum and minimum values of int data ma, mi in a reasonable interval,
Outlier series data da_ex
st->op1->op2->op3->e
st=>start: Draw vertical lines from mi to p_25 and from p_75 to ma
e=>end: Finally, according to the multiplebox parameter, it can decide whether to draw only one median line or multiple lines according to the 5% scale
op1=>operation: Draw a horizontal line with y value equal to mi and a horizontal line with y = ma
op2=>operation: Draw the box in the middle part of the box
op3=>operation: According to the y-axis coordinates provided by the list type parameter da_ex, the method of text annotation is used to draw the abnormal value
Draw a white vertical line from the lowest point to the highest point of the outlier, and set the z-axis parameter to 0, so that the y-axis can cover the abnormal value displayed in the annotation form
st->op1->op2->op3->e
def histobox_plot(ax, data_l,linecolor='black',pointcolor='black',boxcolor='lavender'):
ax.cla()
lenth=len(data_l)
ax.set_xlim((0,lenth+1))
tick_x=range(1,lenth+1)
ax.set_xticks(tick_x)
ax.set_xticklabels([str(i) for i in tick_x])
def normal_his(da,Rec_begin):
his_l=[]
for i in range(len(Rec_begin)-1):
his_l.append(da[(da<Rec_begin[i+1]) & (da>=Rec_begin[i])].count())
his_l.append(da[da>=Rec_begin[len(Rec_begin)-1]].count())
ma_l=max(his_l)
mi_l=min(his_l)
his_nor=[((i-mi_l)/(ma_l-mi_l))*0.55 for i in his_l]
#标准化 [0,1]*0.55
return his_nor
def draw_histobox(ax,mi,ma,p_25,p_50,p_75,da_ex,Rec_begin,his_nor,da_m,da_n,k,linecolor='black',pointcolor='black',boxcolor='lavender'):
ax.vlines(k,mi,ma,color=linecolor,linewidth=1)
ax.hlines(mi,k-0.15,k,color=linecolor,linewidth=1)
ax.hlines(ma,k-0.15,k,color=linecolor,linewidth=1)
ax.hlines(p_50,k-0.25,k,color=linecolor,linewidth=1)
rect = plt.Rectangle((k-0.25,p_25),0.25,p_75-p_25,linewidth=1, edgecolor=linecolor, facecolor='none')
ax.add_patch(rect)
for i in da_ex.values:
ax.text(k,i , "o", fontsize=5,color=pointcolor, verticalalignment="center",horizontalalignment='center')
#ax.scatter(k,i,color='white', marker='o', edgecolors='black')
ax.vlines(k,da_ex.min(),da_ex.max(),color='white',linewidth=1,zorder=0)
for i in range(len(Rec_begin)):
rect = plt.Rectangle((k,Rec_begin[i]),0.15+his_nor[i],(da_m-da_n)/10,linewidth=1,edgecolor=linecolor, facecolor=boxcolor)
ax.add_patch(rect)
def main_box(ax,data,k,linecolor,pointcolor,boxcolor):
da=pd.Series(data).astype(int)
da=da.sort_values(ascending=False)
da.reset_index(drop=True, inplace=True)
p_25,p_50,p_75=myplt.cal_Quantile(da)[:3]
ma,mi=myplt.cal_max_min(da,p_25,p_75)
da_ex=da[(da<(p_25-(p_75-p_25)*1.5)) | (da>(p_25+(p_75-p_25)*1.5))]
#这里改
Rec_begin=np.arange(da.min(),da.max(),(da.max()-da.min())/10)
his_nor=normal_his(da,Rec_begin)
draw_histobox(ax,mi,ma,p_25,p_50,p_75,da_ex,Rec_begin,his_nor,da.max(),da.min(),k,linecolor,pointcolor,boxcolor)
for k in range(lenth):
main_box(ax,data_l[k],k+1,linecolor,pointcolor,boxcolor)Rec_begin：Is list data. The content is the result of a single column of all data da divided into ten equal parts, starting from the minimum value.
The starting Y-axis coordinate used to draw the histogram interval.
normal_his：Input all data da and list data Rec_begin.
The function is to count the number of ten intervals divided by Rec_begin.
Get the his_l and standardize it to [0,0.55].
Get the standardized data list his_nor and return it.
draw_histobox：Draw according to the input parameters.
main_box：The general manager to call various function functions.
st=>start: Empty the canvas and set the X-axis
e=>end: Data input into draw_histobox() for drawing
op1=>operation: The for loop makes every data entered into main_box() as a single-layer list data, and (k+1) is the x-axis scale
op2=>operation: The main_box() first converts list data into single column Series type data, and then calls corresponding functions after sorting
op3=>operation: calculate the quartile int data type p_25, p_50, p_75, series data da3,
The maximum and minimum values of int data ma, mi in a reasonable interval,
Outlier series data da_ex,
The starting Y-axis coordinate recbegin and histogram square length his_nor of histogram interval
st->op1->op2->op3->e
st=>start: Draw a vertical line from mi to ma
e=>end: The his_nor overall + 0.15, making the data set in [0.15,0.7]
Draw
op1=>operation: Draw a horizontal line with y value equal to mi and a horizontal line with y = ma
op2=>operation: Draw the left half of the box diagram box
op3=>operation: According to the y-axis coordinates provided by the list type parameter da_ex, the method of text annotation is used to draw the abnormal value
Draw a white vertical line from the lowest point to the highest point of the outlier, and set the z-axis parameter to 0, so that the y-axis can cover the abnormal value displayed in the annotation form
op4=>operation: Draw the rectangle of histogram according to the bottom coordinate given by Rec_begin and the width coordinate given by his_nor.
st->op1->op2->op3->op4->e
def creative_boxplot(ax, data_l,linecolor='black',pointcolor='black',boxcolor='lavender'):
ax.cla()
lenth=len(data_l)
ax.set_xlim((0,lenth*3+1))#lenth+lenth*2+1
tick_x=range(1,lenth*3+1)
his_x=list(tick_x)[:lenth]
sim_x=[j for j in list(tick_x)[lenth:] if j%2==0]
ax.set_xticks(his_x+sim_x)
ax.set_xticklabels([str(i) for i in his_x]+["zoom"+str(i) for i in his_x])
da_ma,da_mi=myplt.cul_all_max_min(data_l)
step=myplt.cul_step(da_ma,da_mi)
tick_y=list(range(da_mi-(da_mi%step),da_ma+step*2,step))
ax.set_ylim((tick_y[0]-step/6,tick_y[-1]+step/6))
ax.set_yticks(tick_y)
ax.set_yticklabels([str(i) for i in tick_y])
def normal_his(da,Rec_begin):
his_l=[]
for i in range(len(Rec_begin)-1):
his_l.append(da[(da<Rec_begin[i+1]) & (da>=Rec_begin[i])].count())
his_l.append(da[da>=Rec_begin[len(Rec_begin)-1]].count())
ma_l=max(his_l)
mi_l=min(his_l)
his_nor=[((i-mi_l)/(ma_l-mi_l))*0.55 for i in his_l]
#标准化 [0,1]*0.55
return his_nor
def draw_histobox(ax,mi,ma,p_25,p_50,p_75,da_ex,Rec_begin,his_nor,da_m,da_n,k,linecolor='black',pointcolor='black',boxcolor='lavender'):
ax.vlines(k,mi,ma,color=linecolor,linewidth=1)
ax.hlines(mi,k-0.15,k,color=linecolor,linewidth=1)
ax.hlines(ma,k-0.15,k,color=linecolor,linewidth=1)
ax.hlines(p_50,k-0.25,k,color=linecolor,linewidth=1)
rect = plt.Rectangle((k-0.25,p_25),0.25,p_75-p_25,linewidth=1, edgecolor=linecolor, facecolor='none')
ax.add_patch(rect)
for i in da_ex.values:
ax.text(k,i , "o", fontsize=5,color=pointcolor, verticalalignment="center",horizontalalignment='center')
#ax.scatter(k,i,color='white', marker='o', edgecolors='black')
ax.vlines(k,da_ex.min(),da_ex.max(),color='white',linewidth=1,zorder=0)
for i in range(len(Rec_begin)):
rect = plt.Rectangle((k,Rec_begin[i]),0.15+his_nor[i],(da_m-da_n)/10,linewidth=1,edgecolor=linecolor, facecolor=boxcolor)
ax.add_patch(rect)
def draw_y_axis(mi,ma,step,step1,lenth,tick_y,ratio,k):
tick_y1=list(range(mi-(mi%step1),ma+step1*2,step1))
while len(tick_y1)!=len(tick_y):
if (len(tick_y1)<len(tick_y)):
tick_y1.append(tick_y1[-1]+step1)
else:
tick_y1.pop()
ax.vlines(lenth+2*k-1,(tick_y1[0]-step/6)*ratio,(tick_y1[-1]+step/6)*ratio,color='black',linewidth=1)
for i in tick_y1:
ax.hlines(i*ratio,lenth+2*k-1-0.05,lenth+2*k-1,color='black',linewidth=1)
ax.text(lenth+2*k-1-0.05, i*ratio , str(i), fontsize=10, verticalalignment="center",horizontalalignment='right')
def draw_sim_box(ax,mi,ma,p_25,p_50,p_75,ratio,k,lenth,linecolor='black'):
ax.vlines(lenth+2*k,mi*ratio,p_25*ratio,color=linecolor,linewidth=1)
ax.vlines(lenth+2*k,p_75*ratio,ma*ratio,color=linecolor,linewidth=1)
ax.hlines(mi*ratio,lenth+2*k-0.15,lenth+2*k+0.15,color=linecolor,linewidth=1)
ax.hlines(ma*ratio,lenth+2*k-0.15,lenth+2*k+0.15,color=linecolor,linewidth=1)
rect = plt.Rectangle((lenth+2*k-0.25,p_25*ratio),0.5,(p_75-p_25)*ratio,linewidth=1, edgecolor=linecolor, facecolor='none')
ax.add_patch(rect)
ax.hlines(p_50*ratio,lenth+2*k-0.25,lenth+2*k+0.25,color=linecolor,linewidth=3)
def main_box(ax,data,k,linecolor,pointcolor,boxcolor):
da=pd.Series(data).astype(int)
da=da.sort_values(ascending=False)
da.reset_index(drop=True, inplace=True)
p_25,p_50,p_75=myplt.cal_Quantile(da)[:3]
ma,mi=myplt.cal_max_min(da,p_25,p_75)
step1=myplt.cul_step(ma,mi)
ratio=step/step1
da_ex=da[(da<(p_25-(p_75-p_25)*1.5)) | (da>(p_25+(p_75-p_25)*1.5))]
Rec_begin=np.arange(da.min(),da.max(),(da.max()-da.min())/10)
his_nor=normal_his(da,Rec_begin)
draw_histobox(ax,mi,ma,p_25,p_50,p_75,da_ex,Rec_begin,his_nor,da.max(),da.min(),k,linecolor,pointcolor,boxcolor)
draw_y_axis(mi,ma,step,step1,lenth,tick_y,ratio,k)
draw_sim_box(ax,mi,ma,p_25,p_50,p_75,ratio,k,lenth,linecolor)
for k in range(lenth):
main_box(ax,data_l[k],k+1,linecolor,pointcolor,boxcolor)normal_his、draw_histobox：refer to the hisbox code to explain the content in the same location
draw_y_axis：scale the Y-axis of the enlarged view
draw_sim_box：draw the enlarged simple box diagram
main_box：the general manager to call various function functions
st=>start: Empty the canvas and set the X-axis
e=>end: Data input into the draw_histobox() to draw the original image,
draw_y_axis() draws the y-axis coordinates of the enlarged drawing,
draw_sim_box() draws enlarged box diagram
op1=>operation: The maximum value da_ma and the minimum value da_mi of all layers are calculated
So that, it can calculate the step size. Sets the common Y-axis of all boxes
op2=>operation: The for loop makes each data entering main_box() as a single layer list data, The (k+1) is the X-axis scale
op3=>operation: main_box () first converts list data into single column Series data, and then calls corresponding functions after sorting
op4=>operation: The quartile int data type was calculated as p_25,p_50,p_75,and series data type da3,
the maximum and minimum values of int data type ma, mi in a reasonable interval,
The Y-axis of the enlarged drawing is step1,
Magnification ratio of enlarged and original drawings ratio,
Outlier series data da_ ex,
The starting Y-axis coordinate of histogram interval is Rec_begin and histogram grid length is his_nor
st->op1->op2->op3->op4->e
data enter
info_boxplot,True
info_boxplot,False
histobox_plot
creative_boxplot
