Splines give me lockjaw. None of them can be used for extrapolaion, even just 1 step ahead. The complex spline you give is excellent within the data range, but I needed an autocorrelion adjunct to get some forecasts/extrapolations, etc.
By the way, it takes a considerable amount of work to scan the MathCad program & translate it into French with the miserable Glabalink (Lernout & Hauspie) program from a CD.