熊猫的：来自pd.crosstab（）的矩阵

Question

I'm progressively learning pandas, I figured out that pd.crosstab() can do marvels but I've hard time to make it work in that case.

I have a list of objects obj tagged with an int, I want to have the matrix of the objects sharing the same tag (1 if it's the same, 0 else)

| obj | tag |
|-----|-----|
| a   | 0   |
| b   | 2   |
| c   | 1   |
| ... | ... |
| z   | 2   |

->

|     | a | b | c | ... | z |
|-----|---|---|---|-----|---|
| a   | 1 | 0 | 0 | .   | 0 |
| b   | 0 | 1 | 0 | .   | 1 |
| c   | 0 | 0 | 1 | .   | 0 |
| ... | . | . | . | .   | 0 |
| z   | 0 | 1 | 0 | 0   | 1 |

There are some for midables ways to do it, is there one more panda-friendly ?

PS : Tried with pd.crosstab(df.obj, df.obj, values=df.tag, aggfunc=[np.sum]) but NaN filled.

Answer 1

You can use merge with crosstab and DataFrame.rename_axis :

df = df.merge(df, on='tag')
df = pd.crosstab(df.obj_x, df.obj_y).rename_axis(None).rename_axis(None, axis=1)
print (df)
   a  b  c  z
a  1  0  0  0
b  0  1  0  1
c  0  0  1  0
z  0  1  0  1