Engauge quality check

[Ipuch]

Opened by: fmoissenet on Jul 7, 2023

Comments:

fmoissenet (Collaborator) - Jul 7, 2023:

Suggested using the large dataset of McClure et al., which includes both Engauge data and author table data for every plot.
Proposed defining an average Engauge error (possibly RMSE) related to manual data extraction.

Ipuch (Owner) - Jul 7, 2023:

Added the Discussion label.
Questioned whether RMSE or any metric provided by Engauge itself could be used.

Ipuch (Owner) - Jul 7, 2023:

Assigned the issue to fmoissenet.

fmoissenet (Collaborator) - Jul 10, 2023:

Mentioned being accustomed to computing RMSE, R2, and peak-to-peak error in such cases.
Asked for feedback on these metrics.

Ipuch (Owner) - Jul 10, 2023:

Elaborated on the initial comment, inquiring if Engauge provides a precision metric.
Suggested using Mean Absolute Error (MAE) instead of RMSE, expecting the error to be below one, as RMSE tends to minimize values below one.
Agreed that R2 and peak-to-peak error are also fine.
Raised the question of evaluating the consequences on rotation matrix terms, specifically x, y, z axes in global coordinates.

Summary:

The issue discusses using a dataset to evaluate the average error in manual data extraction with Engauge. Metrics like RMSE, MAE, R2, and peak-to-peak error are considered for this purpose. The discussion also touches on the implications for rotation matrix terms.