Similar idea in that features are perturbed but both algorithms are quite different. E.g. sklearn uses scores but featureimpact doesn't; sklearn uses random perturbations but featureimpact uses quantiles.