hstack内存不足

Question

我得到scipy.sparse.csr_matrix对象，其大小为full_sites_sparse.shape (336358, 48371)。我也有尺寸为union_df[hour].shape (336358, 17)的数据框。我想在那里合并并获得另一个(336358, 48388)尺寸，我尝试在其中合并：

full_sites_hour_sparse = np.hstack((full_sites_sparse.A, union_df[hour].values))

和

full_sites_hour_sparse = scipy.sparse.hstack(full_sites_sparse.A, union_df[hour].values)

但是两者都引发内存不足异常。还有其他方法吗？

Answer 1

from scipy.sparse import csr_matrix, hstack, coo_matrix, vstack
t2 = csr_matrix(union_df[hour].values)
diff_n_rows = full_sites_sparse.shape[0] - t2.shape[0]
Xb_new = vstack((t2, csr_matrix((diff_n_rows, t2.shape[1]))))
X = hstack((full_sites_sparse, Xb_new))