Double-hyperbolic undersampling & probabilistic benchmarks

Benchmarks are standards that allow to identify opportunities for improvement among comparable units. The code performs a 2-step estimation of probabilistic benchmarks in noisy data sets: (i) double-hyperbolic undersampling filters the noise of key performance indicators (KPIs), and (ii) relevance vector machines estimate probabilistic benchmarks with the denoised KPIs. The usefulness of the methods is illustrated with an application to a database of nano-finance+.