Question

我有以下df1：

               prueba
 12-03-2018     7
 08-03-2018     1
 06-03-2018     9
 05-03-2018     5

我想让列中的每个值都按最后一个（5）开始，然后用该值减去整个列。然后向上迭代并减去列中的剩余值。对于每次减法，我想生成一个列并生成一个带有每个减法结果的df：

所需的输出将是这样的：

             05-03-2018  06-03-2018  08-03-2018  12-03-2018
 12-03-2018     2           -2           6           0
 08-03-2018    -4           -8           0          NaN
 06-03-2018     4            0           NaN        NaN
 05-03-2018     0           NaN          NaN        NaN

我尝试获得所需的输出，首先是df1和

df2=df1.sort_index(ascending=True)

创建一个空的df：

main_df=pd.DataFrame()

然后迭代列df2中的值并减去df1列

for index, row in df2.iterrows():
    datos=df1-row['pruebas']
    df=pd.DataFrame(data=datos,index=index)
    if main_df.empty:
        main_df= df
    else:
        main_df=main_df.join(df)

print(main_df)

但是以下错误输出：

TypeError: Index(...) must be called with a collection of some kind, '05-03-2018' was passed

Answer 1

您可以使用np.triu，使用数组减法

s=df.prueba.values.astype(float)
s=np.triu((s-s[:,None]).T)
s[np.tril_indices(s.shape[0], -1)]=np.nan
pd.DataFrame(s,columns=df.index,index=df.index).reindex(columns=df.index[::-1])
Out[482]: 
            05-03-2018  06-03-2018  08-03-2018  12-03-2018
12-03-2018         2.0        -2.0         6.0         0.0
08-03-2018        -4.0        -8.0         0.0         NaN
06-03-2018         4.0         0.0         NaN         NaN
05-03-2018         0.0         NaN         NaN         NaN

Answer 2

有点乱，但做了工作：

temp = 0
count = 0
df_new = pd.DataFrame()
for i, v, date in zip(df.index, df["prueba"][::-1], df.index[::-1]):
    print(i,v)
    new_val = df["prueba"] - v
    if count > 0:
        new_val[-count:] = np.nan
    df_new[date] = new_val
    temp += v
    count += 1
df_new

按整列减去列中的每个值

2 个答案: