python - How to transpose a dataset in a csv file?

23
votes

For example, i would like to transform:

Name,Time,Score
Dan,68,20
Suse,42,40
Tracy,50,38

Into:

Name,Dan,Suse,Tracy
Time,68,42,50
Score,20,40,38

EDIT: the original question used the term "transpose" incorrectly.

7
pythoncsvtranspose
What have you tried? Is this homework? WHY do you want to transform it from something usual to something weird? – John Machin
lol. i wish i were back to the times of hw – zr.
I don't think transpose is incorrect here, personally. See en.wikipedia.org/wiki/Transpose – chmullig
@chmullig: ??? matrix transposition is EXACTLY what the OP wants. "pivot" is not the word anyone should be looking for. – John Machin
@chmullig: looks like I missed the "in" in incorrect ... I'm not usually fazed by double negatives; I'm the guy who writes not not x instead of bool(x) :-) – John Machin

7 Answers

41
votes

If the whole file contents fits into memory, you can use

import csv
from itertools import izip
a = izip(*csv.reader(open("input.csv", "rb")))
csv.writer(open("output.csv", "wb")).writerows(a)

You can basically think of zip() and izip() as transpose operations:

a = [(1, 2, 3),
     (4, 5, 6),
     (7, 8, 9)]
zip(*a)
# [(1, 4, 7),
#  (2, 5, 8),
#  (3, 6, 9)]

izip() avoids the immediate copying of the data, but will basically do the same.

16
votes

Transfer from input.csv to output.csv. Pandas can also help.

import pandas as pd
pd.read_csv('input.csv', header=None).T.to_csv('output.csv', header=False, index=False)
5
votes

Same answer of nosklo (all credits to him), but for python3:

from csv import reader, writer 
with open('source.csv') as f, open('destination.csv', 'w') as fw: 
    writer(fw, delimiter=',').writerows(zip(*reader(f, delimiter=',')))
2
votes
from itertools import izip
from csv import reader, writer

with open('source.csv') as f, open('destination.csv', 'w') as fw:
    writer(fw, delimiter=',').writerows(izip(*reader(f, delimiter=',')))
1
votes

If lines is the list of your original text than it should be

for i in range(1,len(lines)):
    lines[i] = lines[i].split(',')

new_lines = []
for i in range(len(lines[0])):
    new_lines.append("%s,%s,%s" % (lines[0][i], lines[1][i], lines[2][i]))

or use csv Python module - http://docs.python.org/library/csv.html

0
votes

The simplest way is:

import numpy as np
import pandas as pd

_mat = pd.read_csv("test.csv")
_mat = _mat[_mat.columns[0:3]].values
_t_mat = np.transpose(_mat)

Result:

  • Input matrix is : [[1 2 3] [4 5 6]]
  • the output is: [[1 4] [2 5] [3 6]]
0
votes

Read the CSV into pandas data frame, pandas has build in function for transpose which can be invoked as below.

import pandas as pd

csv = pd.read_csv("test.csv", skiprows=1)
# use skiprows if you want to skip headers
df_csv = pd.DataFrame(data=csv)
transposed_csv = df_csv.T
print(transposed_csv)