我有一个像这样的数据框,我正在尝试使用Pandas的Pivot重塑我的数据框,使我可以保留原始行中的一些值,同时将重复行变成列并重命名。有时候我有5行重复的行
我一直在尝试,但是我不明白。
import pandas as pd
df = pd.read_csv("C:dummy")
df = df.pivot(index=["ID"], columns=["Zone","PTC"], values=["Zone","PTC"])
# Rename columns and reset the index.
df.columns = [["PTC{}","Zone{}"],.format(c) for c in df.columns]
df.reset_index(inplace=True)
# Drop duplicates
df.drop(["PTC","Zone"], axis=1, inplace=True)
输入
ID Agent OV Zone Value PTC
1 10 26 M1 10 100
2 26.5 8 M2 50 95
2 26.5 8 M1 6 5
3 4.5 6 M3 4 40
3 4.5 6 M4 6 60
4 1.2 0.8 M1 8 100
5 2 0.4 M1 6 10
5 2 0.4 M2 41 86
5 2 0.4 M4 2 4
输出
ID Agent OV Zone1 Value1 PTC1 Zone2 Value2 PTC2 Zone3 Value3 PTC3
1 10 26 M_1 10 100 0 0 0 0 0 0
2 26.5 8 M_2 50 95 M_1 6 5 0 0 0
3 4.5 6 M_3 4 40 M_4 6 60 0 0 0
4 1.2 0.8 M_1 8 100 0 0 0 0 0 0
5 2 0.4 M_1 6 10 M_2 41 86 M_4 2 4
答案 0 :(得分:2)
将cumcount
用于计数组,使用set_index
由unstack
创建MultiIndex
,并最后平整列的值:
g = df.groupby(["ID","Agent", "OV"]).cumcount().add(1)
df = df.set_index(["ID","Agent","OV", g]).unstack(fill_value=0).sort_index(axis=1, level=1)
df.columns = ["{}{}".format(a, b) for a, b in df.columns]
df = df.reset_index()
print (df)
ID Agent OV Zone1 Value1 PTC1 Zone2 Value2 PTC2 Zone3 Value3 PTC3
0 1 10.0 26.0 M1 10 100 0 0 0 0 0 0
1 2 26.5 8.0 M2 50 95 M1 6 5 0 0 0
2 3 4.5 6.0 M3 4 40 M4 6 60 0 0 0
3 4 1.2 0.8 M1 8 100 0 0 0 0 0 0
4 5 2.0 0.4 M1 6 10 M2 41 86 M4 2 4
如果只想替换为0
个数字列:
g = df.groupby(["ID","Agent"]).cumcount().add(1)
df = df.set_index(["ID","Agent","OV", g]).unstack().sort_index(axis=1, level=1)
idx = pd.IndexSlice
df.loc[:, idx[['Value','PTC']]] = df.loc[:, idx[['Value','PTC']]].fillna(0).astype(int)
df.columns = ["{}{}".format(a, b) for a, b in df.columns]
df = df.fillna('').reset_index()
print (df)
ID Agent OV Zone1 Value1 PTC1 Zone2 Value2 PTC2 Zone3 Value3 PTC3
0 1 10.0 26.0 M1 10 100 0 0 0 0
1 2 26.5 8.0 M2 50 95 M1 6 5 0 0
2 3 4.5 6.0 M3 4 40 M4 6 60 0 0
3 4 1.2 0.8 M1 8 100 0 0 0 0
4 5 2.0 0.4 M1 6 10 M2 41 86 M4 2 4
答案 1 :(得分:1)
您可以使用cumcount
创建帮助键,然后对多个索引进行平坦化({:您可以在末尾添加fillna(0),我没有添加它,原因是不认为Zone值0是正确的)
unstack