文件方式实现完整的英文词频统计实例

fo=open('test.txt','r')
a=fo.read()
exc={'the','a','to','of','and','in','that','on',''}
a=a.lower()
for i in ',.':
    a=a.replace(i,' ')
words=a.split(' ')
print(words)
dic={}
keys=set(words)
keys=keys-exc


print (keys)

for i in keys:
    dic[i]=words.count(i)
print(dic)
wc=list(dic.items())
wc.sort(key=lambda x:x[1],reverse=True)
print(dic)
for i in range(20):
     print(wc[i])
fo.close()
     

运行结果：