具有以pandas数据集列作为行的数据透视表的有效方法是什么？

Question

Let's take as an example the following dataset:

    make    address all     3d   our    over    length_total    y
0   0.0     0.64    0.64    0.0  0.32   0.0     278             1
1   0.21    0.28    0.5     0.0  0.14   0.28    1028            1
2   0.06    0.0     0.71    0.0  1.23   0.19    2259            1
3   0.15    0.0     0.46    0.1  0.61   0.0     1257            1
4   0.06    0.12    0.77    0.0  0.19   0.32    749             1
5   0.0     0.0     0.0     0.0  0.0    0.0     21              1
6   0.0     0.0     0.25    0.0  0.38   0.25    184             1
7   0.0     0.69    0.34    0.0  0.34   0.0     261             1
8   0.0     0.0     0.0     0.0  0.9    0.0     25              1
9   0.0     0.0     1.42    0.0  0.71   0.35    205             1
10  0.0     0.0     0.0     0.0  0.0    0.0     23              0
11  0.48    0.0     0.0     0.0  0.48   0.0     37              0
12  0.12    0.0     0.25    0.0  0.0    0.0     491             0
13  0.08    0.08    0.25    0.2  0.0    0.25    807             0
14  0.0     0.0     0.0     0.0  0.0    0.0     38              0
15  0.24    0.0     0.12    0.0  0.0    0.12    227             0   
16  0.0     0.0     0.0     0.0  0.75   0.0     77              0
17  0.1     0.0     0.21    0.0  0.0    0.0     571             0
18  0.51    0.0     0.0     0.0  0.0    0.0     74              0
19  0.3     0.0     0.15    0.0  0.0    0.15    155             0

I want to get pivot-table from the previous dataset, in which the columns (make, address all, 3d, our, over, length_total) will have their mean values processed by the column y. The following table is the expected result:

                    y   
                    1   0
make            0.048   0.183
address         0.173   0.008
all             0.509   0.098
3d              0.01    0.02
our             0.482   0.123
over            0.139   0.052
length_total    626.7   250

Is it possible to get the desired result through pivot_table method from pandas.data object? If so, how?

Is there a more effective way to do this?

Answer 1

Some people like using stack or unstack , but I prefer good ol' pd.melt to "flatten" or "unpivot" a frame:

>>> df_m = pd.melt(df, id_vars="y")
>>> df_m.pivot_table(index="variable", columns="y")
                value         
y                   0        1
variable                      
3d              0.020    0.010
address         0.008    0.173
all             0.098    0.509
length_total  250.000  626.700
make            0.183    0.048
our             0.123    0.482
over            0.052    0.139

(If you want to preserve the original column order as the new row order, you can use .loc to index into this, something like df2.loc[df.columns].dropna() ).

---

Melting does the flattening, and preserves y as a column, putting the old column names as a new column called "variable" (which can be changed if you like):

>>> pd.melt(df, id_vars="y").head()
   y variable  value
0  1     make   0.00
1  1     make   0.21
2  1     make   0.06
3  1     make   0.15
4  1     make   0.06

After that we can call pivot_table as we would ordinarily.