Kalman-and-Bayesian-Filters-in-Python
Kalman-and-Bayesian-Filters-in-Python copied to clipboard
rlabbe
Accuracy vs Precision vs Resolution issues
Here it is said that the narrowly distributed temperature readings represent a more accurate thermometer, when I think it is intended to mean a more precise one. The book previously established that we will assume no systemic bias, or inaccuracy. This occurs in other places before this, and perhaps in future sections of the book. If I track down more instances of this I'll add them as comments to this issue.
https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/blob/9e3d2f6ed023d937587cf2ef2ecfbf7afc3d8054/03-Gaussians.ipynb#L1188
https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/blob/9e3d2f6ed023d937587cf2ef2ecfbf7afc3d8054/03-Gaussians.ipynb#L1003
Here is another instance. I think in context that "precise" is meant instead of "accurate". While it is a true statement that no thermometer is perfectly accurate, the variance from reading to reading that is being alluded to here is a precision issue, not necessarily an accuracy one.
Here "precision" is used when I think "resolution" should be used. "precision" is meant to convey how tightly together readings are clustered, where "resolution" is the minimum delta that a sensor can display.
https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/blob/9e3d2f6ed023d937587cf2ef2ecfbf7afc3d8054/03-Gaussians.ipynb#L1019
https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/blob/9e3d2f6ed023d937587cf2ef2ecfbf7afc3d8054/04-One-Dimensional-Kalman-Filters.ipynb#L639
https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/blob/9e3d2f6ed023d937587cf2ef2ecfbf7afc3d8054/04-One-Dimensional-Kalman-Filters.ipynb#L645
I think "accurate" should be replaced with "precise" in these two instances as well.
https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/blob/9e3d2f6ed023d937587cf2ef2ecfbf7afc3d8054/05-Multivariate-Gaussians.ipynb#L852
Perhaps instead of "error" it should be "confidence" or "uncertainty"