问题描述
我想像时钟一样周期性地旋转行,但我希望每一行 会根据“ n_roll”列的不同而旋转
所以如果我有那个df
data={"col1":[2,3,4,5],"col2":[4,2,6],"col3":[7,6,9,11],"col4":[14,11,22,8],"name":["A","A","V","A"],"n_roll":[1,3]}
df=pd.DataFrame.from_dict(data)
df
所以我希望它看起来像这样
data={"col1":[14,"col2":[2,"col3":[4,"col4":[7,3]}
df=pd.DataFrame.from_dict(data)
df
也许是这样的: coll_to_roll = [“ col1”,“ col2”,“ col3”,“ col4”]
df[coll_to_roll] = np.roll(df[coll_to_roll],1,df["n_roll"])
解决方法
您可以通过将DataFrame和column转换为numpy数组来重用现有功能:
coll_to_roll=["col1","col2","col3","col4"]
from skimage.util.shape import view_as_windows as viewW
#https://stackoverflow.com/a/51613442
def strided_indexing_roll(a,r):
# Concatenate with sliced to cover all rolls
a_ext = np.concatenate((a,a[:,:-1]),axis=1)
# Get sliding windows; use advanced-indexing to select appropriate ones
n = a.shape[1]
return viewW(a_ext,(1,n))[np.arange(len(r)),(n-r)%n,0]
df[coll_to_roll]=strided_indexing_roll(df[coll_to_roll].to_numpy(),df["n_roll"].to_numpy())
print (df)
col1 col2 col3 col4 name n_roll
0 14 2 4 7 A 1
1 6 11 3 2 A 2
2 9 22 4 4 V 2
3 6 11 8 5 A 3