Here’s that side‑by‑side view brought to life 📊 — you can see the record (blue) and target (green) spectra tracing along the period axis, with each red dashed segment marking the gap at that point. The little ± numbers are the exact differences in cm/s², so you can instantly spot where the record overshoots or undershoots the target.
In our example, the biggest miss is at 0.40 s (–20 cm/s²), while the rest hover around ±10. That’s why the overall RMSE is modest, and after normalization, you get that tidy ~0.05 score. It’s a nice visual confirmation of what the math told us — the fit is pretty tight, with only one standout deviation.
If you’d like, we could layer on the percent error curve as a third line, so you can see the relative mismatch at each period without doing the mental division. That can be handy when comparing multiple records against the same target.
GRAFICO ALLEGATO.
"Data speak for themselves" -Reverend Thomas Bayes 1702-1761 P(Ai|E)=(P(E|Ai)P(Ai))/P(E)
