Visual Diagnostics for Constrained Optimisation with Application to Guided Tours

A guided tour helps to visualise high-dimensional data by showing low-dimensional projections along a projection pursuit optimisation path. Projection is generalisation of principal component analysis, in the sense that different indexes are used define interestingness projected data. While much work has been done developing new literature, less on understanding optimisation. Index functions can be noisy, might have multiple local maxima as well an optimal maximum, and constrained generate orthonormal frames, which complicates optimization. In addition, primarily for exploratory finding also useful. The especially useful exploration, because it conducts geodesic interpolation connecting steps shows how changes approached. This provides visual diagnostics examining choice procedure, based provision object collects information throughout It helped diagnose fix several problems with tour. more broadly diagnosing optimisers, comparing their performance. implemented R package, ferrn.