Could be useful for compute_betas in IWLS to find a sparse-dennse flexible way of using the more robust scipy.linalg.solve method rather than directly computing the inverse of X'X, which would then match how we are computing the GWR estimators in compute_beta_gwr.
Could be useful for
compute_betasin IWLS to find a sparse-dennse flexible way of using the more robustscipy.linalg.solvemethod rather than directly computing the inverse of X'X, which would then match how we are computing the GWR estimators incompute_beta_gwr.