解析 csv 文件，将其列更改为行，将行更改为列

Question

我有一个 CSV 文件，其中第一行是日期，第一列是扇区。像这样。

Date,7/2/2007,7/3/2007,7/5/2007,7/6/2007,7/9/2007
A,0,1,3,2,0
AA,23,423,2,0,0
AAL,34,23,5,0,234
AGCG,234,0,9,234,23
XL,0,65,34,34,34

所以现在我想准备另一个文件，就像

Date,Sector
7/2/2007,AA
7/2/2007,AAL
7/2/2007,AGCG
7/3/2007,A
7/3/2007,AA
7/3/2007,AAL
7/3/2007,XL
...

背后的逻辑是，我希望每个日期的扇区不为 0。

到目前为止我尝试过的代码是：导入 csv,sys 从集合导入 defaultdict

dd = defaultdict(list)
dateList = []
header = False

def createFile(di):
    ff = open("cum_file.csv","w")
    csvwriter = csv.writer(ff)
    row = []
    for d,t in di.iteritems():
        for tt in t:
            print tt,d
            row = [tt,d]
            csvwriter.writerow(row)
            del row[:]

#with open("./data/StrategyAcctValue-Daily.csv") as f:
with open("./try/test.csv") as f:
    reader = csv.reader(f,delimiter=",")
    for line in reader:
        col1 = True
        if header:
            #sys.exit()
            for eachCol in line:
                if col1:
                    col1 = False
                    tkr = eachCol
                elif eachCol != '0':
                    tkrIndex = line.index(eachCol)
                    tickerDate = dateList[tkrIndex - 1]
                    dd[tickerDate].append(tkr)
                else:
                    continue
            #print dd
            #createFile(dd)
            #sys.exit()
        else:
            header = True
            for eachCol in line:
            #   print line.index(eachCol)
            #   continue
                if col1:
                    col1 = False
                    tkr = eachCol
                else:
                    dd[eachCol] = []
                    dateList.append(eachCol)
            print dateList
    print dd
    createFile(dd)

这给出了如下输出：

A 7/3/2007
AA 7/3/2007
AAL 7/3/2007
XL 7/3/2007
A 7/6/2007
AAL 7/9/2007
AGCG 7/9/2007
AA 7/2/2007
AAL 7/2/2007
AGCG 7/2/2007
AGCG 7/2/2007
A 7/5/2007
AA 7/5/2007
AAL 7/5/2007
AGCG 7/5/2007
XL 7/5/2007
XL 7/5/2007
XL 7/5/2007

我找不到哪里出错了。

Answer 1

with open("test.csv", 'r') as f:
    data = list(zip(*(line.rstrip("\n").split(',') for line in f)))
    temp = dict(enumerate(data[0][1:]))
    for key, *values in data[1:]:
        for index, value in enumerate(values):
            if value != '0':
                print(key, temp[index])

7/2/2007 AA
7/2/2007 AAL
7/2/2007 AGCG
7/3/2007 A
7/3/2007 AA
7/3/2007 AAL
7/3/2007 XL
7/5/2007 A
7/5/2007 AA
7/5/2007 AAL
7/5/2007 AGCG
7/5/2007 XL
7/6/2007 A
7/6/2007 AGCG
7/6/2007 XL
7/9/2007 AAL
7/9/2007 AGCG
7/9/2007 XL

Answer 2

使用 numpy，您可以将 csv 数据转换为数组（矩阵）。然后你可以迭代转置矩阵。

import numpy as np

data = np.matrix([['Date','7/2/2007','7/3/2007','7/5/2007','7/6/2007','7/9/2007'],
['A',0,1,3,2,0],
['AA',23,423,2,0,0],
['AAL',34,23,5,0,234],
['AGCG',234,0,9,234,23],
['XL',0,65,34,34,34]])

#get the index of the places in a row (from header)
#data.T is transposed matrix
index = data.T[0]
# you iterate over the dates (row of the transposed matrix) skipping the header row
for date in data.T[1:]:
    # get the non-zero element of the row, get the correpond place (using the index)
    # range begin at 1 to avoid the first column (contains date)
    for place in [index[0,i] for i in range(1,date.shape[1]) if date[0,i] != 0]:
        print(date[0,0], place)

Answer 3

我知道问题出在哪里，我正在回答我的问题，并对更改发表评论。

import csv,sys
from collections import defaultdict

dd = defaultdict(list)
dateList = []
header = False

def createFile(di):
    ff = open("cum_file.csv","w")
    csvwriter = csv.writer(ff)
    row = []
    for d,t in di.iteritems():
        for tt in t:
            print tt,d
            row = [tt,d]
            csvwriter.writerow(row)
            del row[:]

with open("./try/test.csv") as f:
    reader = csv.reader(f,delimiter=",")
    for line in reader:
        col1 = True
        if header:
            #sys.exit()
            for eachCol in line:
                if col1:
                    col1 = False
                    tkr = eachCol
                elif eachCol != '0':
                    tkrIndex = line.index(eachCol) #in case of duplicate non - zero values, it is returing the index of first one
                    tickerDate = dateList[tkrIndex - 1]
                    dd[tickerDate].append(tkr)
                    line[tkrIndex] = '' # make the entry blank os duplicate issue will not occur
                else:
                    continue
            #print dd
            #createFile(dd)
            #sys.exit()
        else:
            header = True
            for eachCol in line:
            #   print line.index(eachCol)
            #   continue
                if col1:
                    col1 = False
                    tkr = eachCol
                else:
                    dd[eachCol] = []
                    dateList.append(eachCol)
            print dateList
    print dd
    createFile(dd)

感谢社区抽出宝贵时间。

Answer 4

import pandas as pd

df = pd.read_csv("input.csv")
df1 = pd.DataFrame(columns=["Date", "Sector"])

for i, row in df.iterrows():
    dict_ = dict(row)
    days = [key for key, value in dict_.items() if value != 0]
    days.remove('Date')
    for day in days:
        df1.loc[len(df1)] = [day, dict_["Date"]]

# df1.sort_values(by='Date') // to sort by date
df1.to_csv("output.csv", index=False)

Output.csv

Date,Sector
7/6/2007,A
7/3/2007,A
7/5/2007,A
7/3/2007,AA
7/2/2007,AA
7/5/2007,AA
7/9/2007,AAL
7/3/2007,AAL
7/2/2007,AAL
7/5/2007,AAL
7/9/2007,AGCG
7/6/2007,AGCG
7/2/2007,AGCG
7/5/2007,AGCG
7/9/2007,XL
7/6/2007,XL
7/3/2007,XL
7/5/2007,XL

解析 csv 文件，将其列更改为行，将行更改为列

Parsing a csv file, changing its columns into rows and rows into columns

python

csv

defaultdict