电影样题的数据清洗

# -*- coding: utf-8 -*-

#coding:utf-8

import numpy as np

import pandas as pd

import string

df20105 = pd.read_csv('fbr_d_n_b.csv', delimiter = ',' , names = ['date','film','BoxOfficeReturn'])

df20105 = df20105 .dropna() #去掉NAN数据

# print df20105

val20105 = df20105.values

# print val20105

lst00 = []

for i in range(0,len(val20105)):

x = val20105[i][0] #请填入4位数字 取出日期

lx = x. split('-')

lst00.append(lx[0]+lx[1]+lx[2])

df20105.index=lst00

# print df20105.index

df20105 = df20105. drop(['date'], axis = 1) #删除原来日期的那一列

# df20105.to_csv('a.csv',index='True',encoding='utf-8')

# print type(val20105)

# print df20105[(df20105.index ==' 20100523')&(df20105.BoxOfficeReturn>10000)]

# print df20105[(df20105.index == ' 20100516') & (df20105.film == '让子弹飞')]

# rs1=df20105['BoxOfficeReturn']. groupby([df20105.index,df20105['film']]).mean()

# print rs1

# df = df20105.mean(columns = 'BoxOfficeReturn')

# print df

# '''

# print df20105.columns

# print df20105.describe

# print df20105

# '''

#df20105 = df20105.sort_index()

#print df20105

# print type(val20105)

# df = df20105[(df20105.index == 20100523)&(df20105["BoxOfficeReturn"]>10000)]

# df = df20105[(df20105.BoxOfficeReturn>10000)]

# df = df20105[(df20105.index == 20100516) & (df20105.film == '让子弹飞')]

# print df

# 将日期转换成周数

lsa=df20105.index

lsb=[]

lsyct=[0,31,28,31,30,31,30,31,31,30,31,30,31]

sumdays=0

t=0

y=0

m=0

d=0

for da in lsa:

y=string.atoi(da)/10000

m=string.atoi(da)/100%100

d=string.atoi(da)%100

sumdays=0

for t in lsyct[0:m]:

sumdays += t

sumdays=(sumdays+d-2+((y%4==0 and y%100!=0 or y%400==0) and m > 2))/7

lsb.append("第"+`sumdays`+"周")

df20105.index=lsb

df20105n=df20105.reindex() #50ans： df2015n.reindex()

print df20105n #

# df20105n=df20105n.sort_index(by=None) #50ans： index

# rs1=df20105n['BoxOfficeReturn'].groupby([df20105n.index,df20105n['film']]).mean()

# rs2=df20105n['BoxOfficeReturn'].groupby([df20105n.index,df20105n['film']]).sum()

# print rs1 #平均

# print rs2 #求和

#57ans： index

#58ans： film