Just out of curiosity, was there ever an attempt to use the data from multiple shuttle runs and rather than fit a line thorugh each run, do a partial least squares model across all the runs?
“Just out of curiosity, was there ever an attempt to use the data from multiple shuttle runs and rather than fit a line thorugh each run, do a partial least squares model across all the runs?”
That’s what we did — which is why I analogized it to the stock market. You could get a good average fit for prior data but it was always useless for predicting how the next flight would go. Too many variables affecting thing — inclination, altitude, sun angle, atmospheric density, landing time, ascending vs descending node landing, and probably a bunch of other stuff that we did not even know about.
It looked pretty, sounded impressive — and was absolutely fricking useless.