API：How can I apply a function on rolling DataFrames

[zhangyw49]

well, my function's input is ndarray-like and it's output is a scaler. The question is I have no method to apply this function on rolling object.

from sklearn.decomposition import PCA

df = pd.DataFrame({"A": [1, 2, 3, 4, 5, 6, 7, 8], "B": [2, 3, 4, 5, 6, 7, 8, 9]})

def first_variance(X):
    pca = PCA(n_components=1)
    pca.fit(X)
    return pca.explained_variance_

df.rolling(4).apply(first_variance)

The domo is incorrect because the "apply" method only can apply by index or column.

[jreback]

this is a duplicate of #15095

[qyqyq]

Faced with the same problem, then what the implementation should be now? It seems that recent API of df.rolling() still has no para with respect to table, rather than a row or column.