CODEWITHSUNDEEP
pythonpandasadjacency-matrix
Clemclem
Clemclem

Reputation: 73

convert Pandas dataframe into adjacency matrix

I have a Pandas dataframe (930 rows × 50 columns) that looks like this:

index Keyword A Keyword B Keyword c
Page 1 1 3 1
Page 2 4 0 2
Page 3 0 1 1

I would like to convert it into an adjacency Matrix / Weighted Graph, where each Keyword is a node. The weight would be the sum of combinations between each keywords.

The result would be something along these lines:

Keyword A Keyword B Keyword C
Keyword A 0 3 8
Keyword B 3 0 4
Keyword C 8 4 0

Upvotes: 5

Views: 1338

Answers (1)

orlp
orlp

Reputation: 117701

The solution is deceptively simple:

adj = df.T @ df
np.fill_diagonal(adj.values, 0)

E.g.:

>>> df = pd.DataFrame([[1, 1, 3, 1], [2, 4, 0, 2], [3, 0, 1, 1]],
                      columns=["index", "A", "B", "C"]).set_index("index")
>>> df
       A  B  C
index
1      1  3  1
2      4  0  2
3      0  1  1
>>> adj = df.T @ df
>>> np.fill_diagonal(adj.values, 0)
>>> adj
   A  B  C
A  0  3  9
B  3  0  4
C  9  4  0

Upvotes: 4

Related Questions