另一個更快的解決方案:從第一列
import pandas as pd
import numpy as np
import io
temp=u"""P:,14142,GREEN,800,9;R:,14597,7,y,NaN,Correct;P:,15605,#E5DC22,800,9;R:,16108,7,f,NaN,Correct;P:,17115,GREEN,100,9;R:,17548,7,y,NaN,Correct;P:,18552,#E5DC22,100,9;R:,18972,7,f,NaN,Correct;P:,19979,GREEN,800,9;R:,20379,7,y,NaN,Correct;P:,21387,#E5DC22,800,9;R:,21733,7,f,NaN,Correct;P:,22740,RED,100,9;R:,23139,7,y,NaN,False;P:,24147,BLUE,100,9;R:,24547,7,f,NaN,False;P:,25555,RED,800,9;R:,26043,7,b,NaN,Correct;P:,27051,BLUE,800,9;"""
#after testing replace io.StringIO(temp) to filename
df = pd.read_csv(io.StringIO(temp), sep=':', header=None, lineterminator=';')
print (df)
0 1
0 P ,14142,GREEN,800,9
1 R ,14597,7,y,NaN,Correct
2 P ,15605,#E5DC22,800,9
3 R ,16108,7,f,NaN,Correct
4 P ,17115,GREEN,100,9
5 R ,17548,7,y,NaN,Correct
6 P ,18552,#E5DC22,100,9
7 R ,18972,7,f,NaN,Correct
8 P ,19979,GREEN,800,9
9 R ,20379,7,y,NaN,Correct
10 P ,21387,#E5DC22,800,9
11 R ,21733,7,f,NaN,Correct
12 P ,22740,RED,100,9
13 R ,23139,7,y,NaN,False
14 P ,24147,BLUE,100,9
15 R ,24547,7,f,NaN,False
16 P ,25555,RED,800,9
17 R ,26043,7,b,NaN,Correct
18 P ,27051,BLUE,800,9
首先set_index
索引,然後通過strip
和除去triling ,
通過str.split
創建DataFrame
。最後需要補充0
列名和reset_index
:
df1 = df.set_index(0)[1].str.strip(',').str.split(',', expand=True)
df1.columns = df1.columns + 1
df1.reset_index(inplace=True)
print (df1)
0 1 2 3 4 5
0 P 14142 GREEN 800 9 None
1 R 14597 7 y NaN Correct
2 P 15605 #E5DC22 800 9 None
3 R 16108 7 f NaN Correct
4 P 17115 GREEN 100 9 None
5 R 17548 7 y NaN Correct
6 P 18552 #E5DC22 100 9 None
7 R 18972 7 f NaN Correct
8 P 19979 GREEN 800 9 None
9 R 20379 7 y NaN Correct
10 P 21387 #E5DC22 800 9 None
11 R 21733 7 f NaN Correct
12 P 22740 RED 100 9 None
13 R 23139 7 y NaN False
14 P 24147 BLUE 100 9 None
15 R 24547 7 f NaN False
16 P 25555 RED 800 9 None
17 R 26043 7 b NaN Correct
18 P 27051 BLUE 800 9 None
時序:
def jez(df):
df1 = df.set_index(0)[1].str.strip(',').str.split(',', expand=True)
df1.columns = df1.columns + 1
df1.reset_index(inplace=True)
return (df1)
print (jez(df))
In [310]: %timeit (pd.concat([df[[0]], df[1].str.split(',').apply(pd.Series).iloc[:, 1: 6]], axis=1))
100 loops, best of 3: 4.85 ms per loop
In [311]: %timeit (jez(df))
1000 loops, best of 3: 1.61 ms per loop
你應該格式化你的代碼,以使問題可讀...... – IanS
請有關如何格式化代碼讀取。 – wander95