Visual analysis [data analysis] IMDb Top250 crawling data

Watercress Top250 URL previously crawling to IMDb simple visualization: CSV format data list saved as

Import data ready

#!-*- coding:utf-8 -*-
import pandas as pd
import numpy as np
import matplotlib.pylab as plt
import re
from numpy import rank
from builtins import map
from datashape.coretypes import Map
#http://www.jianshu.com/p/0a76c94e9db7 参考了简书上的饼状图教程
#切换工作目录，IPython运行%pylab
Movie=pd.read_csv('./doubanmovietop.csv') #数据读取

Check the header

Movie.head()

	Unnamed: 0	title	info	rating_num	comment_num	daoyan	date	guojia	juqing
0	1	The Shawshank Redemption	People want freedom.	9.7	1682392	Director: Frank Darabont Frank Darabont Starring: Tim Robbins Tim Robb ...	1994	United States	Crime drama
1	2	Farewell My Concubine	Absolute beauty.	9.6	1244650	Director: Kaige Chen Kaige Chen Cast: Leslie Cheung, Leslie Cheung / Zhang Fengyi ...	1993	Mainland China Hong Kong, China	Homosexual love story
2	3	Forrest Gump	An American modern history.	9.5	1301770	Director: Robert Zemeckis Robert Zemeckis Starring: Tom Hanks Tom Han ...	1994	United States	Love story
3	4	The killer is not too cold	Undead and Lolita story have to say.	9.4	1495321	Director: Luc Besson Luc Besson Starring: Jean Reno Jean Reno / Natalie wave ...	1994	France	Action crime drama
4	5	beautiful life	The most beautiful lies.	9.5	760464	Director: Roberto Benigni is Roberto Benigni Starring: Roberto Bernini Roberto ...	1997	Italy	Drama Comedy Romance War

IMDb Top250 score distribution pie chart Code:

#Rating pie
Rating=Movie['rating_num']
bins=[8,8.5,9,9.5,10]  #分区(0,8],(8,8.5]....
rat_cut=pd.cut(Rating,bins=bins)
rat_class=rat_cut.value_counts()  #统计区间个数
rat_pct=rat_class/rat_class.sum()*100  #计算百分比
rat_arr_pct=np.array(rat_pct)#将series格式转成array，为了避免pie中出现name
f1=plt.figure(figsize=(9,9))
plt.title('DoubanMovieTop250\nRatingDistributin(0~10)')
plt.pie(rat_arr_pct,labels=rat_pct.index,colors=['r','g','b','c'],autopct='%.2f%%',startangle=75,explode=[0.05]*4)  #autopct属性显示百分比的值
plt.savefig('MovieTop250.RatingDistributin(0~10).png')
f1.show()
#explode:将某部分爆炸出来， 使用括号，将第一块分割出来，数值的大小是分割出来的与其他两块的间隙
#labeldistance，文本的位置离远点有多远，1.1指1.1倍半径的位置
#autopct，圆里面的文本格式，%3.1f%%表示小数有三位，整数有一位的浮点数
#shadow，饼是否有阴影
#startangle，起始角度，0，表示从0开始逆时针转，为第一块。一般选择从90度开始比较好看
#pctdistance，百分比的text离圆心的距离
#patches, l_texts, p_texts，为了得到饼图的返回值，p_texts饼图内部文本的，l_texts饼图外label的文本

Film produced in the distribution of the pie:

#year pie
year=Movie['date']
for i in year.index:
    if len(year[i])>4:
        year.drop(i,inplace=True)  #  year.drop(i,inplace=True)  去除多个年代的特例,inplace重要，修改改变原值
year=year.astype(int)
bins=np.linspace(min(year)-1,max(year)+1,10).astype(int)  #产生区间,bins一般为(,]的，所以+1
year_cut=pd.cut(year,bins=bins)
year_class=year_cut.value_counts()
year_pct=year_class/year_class.sum()*100
year_arr_pct=np.array(year_pct)
color=['b', 'g', 'r', 'c', 'm', 'y', (0.2,0.5,0.7), (0.6,0.5,0.7),(0.2,0.7,0.1)]  #RGB 0-1之间的tuple
f2=plt.figure(figsize=(9,9))
patches,out_text,in_text=plt.pie(year_arr_pct,labels=year_pct.index,colors=color,autopct='%.2f%%',explode=[0.05]*9,startangle=30)
plt.title('MovieTop250\nYears Distribution')
f2.show()
# plt.savefig('MovieTop250_YearsDistribution.png')
 

IMDb Top250, the film ranked number Evaluation & movie ratings scatter plot:

#评价人数 
rank=np.array(Movie.index,dtype=int)+1 #index start from 0 
Movie['0']=rank
f3=plt.figure(3,figsize=(12,10))
plt.scatter(x=Movie['0'],y=Movie['comment_num'],c=Movie['rating_num'],s=80)
plt.title('Douban Movie\nRank and Rating People by Rating',fontsize=20)
plt.xlabel('Rank',fontsize=15)
plt.ylabel('Rating People',fontsize=15)
plt.axis([-5,255,0,750000])  #x轴坐标范围
plt.colorbar()  #显示colorbar
plt.savefig('DoubanMovie_Rank_and_RatingPeople_by_Rating.png')
plt.show()

By National Film Classification bar chart :

#!-*- coding:utf-8 -*-
import pandas as pd
import numpy as np
import matplotlib.pylab as plt
from matplotlib.font_manager import FontProperties  #fontproperties的模块，pyde自动添加的，好评

Movie=pd.read_csv('./doubanmovietop.csv',encoding='utf-8')
country_iter=(set(x.split(' ')) for x in Movie['guojia']) #generator生成器，分解字符串
countries=sorted(set.union(*country_iter)) #Return the union of sets as a new set.
#*country_iter:This works for any iterable of iterables.
df=pd.DataFrame(np.zeros((len(Movie),len(countries))),columns=countries)#创建一个0DataFrame，np.zeros()内为要tuple
for i,gen in enumerate(Movie['guojia']): 
    df.ix[i,gen.split(' ')]=1  #第i条数据的country置为1
num_of_country=df.sum() 
# print(num_of_country)
num_of_country[4]=num_of_country[1]+num_of_country[2]+num_of_country[4]  #(1964中国大陆中国大陆重映)和中国大陆合并
# num_of_country.pop('中国')
# print(num_of_country)
num_of_country.sort_values(inplace=True,ascending=False)
f1=plt.figure()
for i,gen in enumerate(num_of_country[:10]):
    plt.bar(i,gen)   #i为bar的起始横坐标，gen为纵坐标，宽度默认
names=list(num_of_country.index)
plt.xticks(np.arange(10),names,fontproperties='SimHei',rotation =60)  #在图中显示中文字符要加上fontproperties='SimHei'
plt.ylabel('Movie Number')
plt.title('Douban Movie\nMovie Distribution by Countries')
# plt.savefig('Movie_Distribution_by_Countries.png')
f1.show()
 #因为有些影片为多国合作的，也算各自国家的吧。
#过滤了很多只有一两部的国家，果然还是美帝有金坷垃，亩产一万八

Genres distribution histogram:

genre_iter=(set(x.split(' ')) for x in Movie['juqing'])
genre=sorted(set.union(*genre_iter))
frame=pd.DataFrame(np.zeros((len(Movie),len(genre))),columns=genre)
for i,gen in enumerate(Movie['juqing']):
    frame.ix[i,gen.split(' ')]=1
genre_sum=frame.sum()
genre_sum.sort_values(inplace=True,ascending=False)
f2=plt.figure(2)
'''for i,gen in enumerate(genre_sum[:8]):
    plt.bar(i,gen)
names=list(genre_sum.index)
plt.xticks(np.arange(8)+0.4,names,fontproperties='SimHei')
plt.show()'''
#改进的方法
p2=plt.bar(np.arange(8),genre_sum.values[:8],align='center') #p2包含8个元素，每个对应一个bar
names=list(genre_sum.index)
plt.xticks(np.arange(8),names,fontproperties='SimHei')
plt.legend((p2[0],),('MovieNumber',)) #只有一个元素的tuple应写成(ele,)
plt.ylabel('Movie Number')
plt.title('Douban Movie\nDistribution by Genre')
#plt.savefig('Movie_Distribution_by_Genre.png')
plt.show()

For more data, please reply back No public concern: IMDb
you can get -