Observations marked as discrepant are excluded from data analysis but need to be removed from the plots or changed to a different series or otherwise assigned a different color.
Please don't exclude them from the plot entirely. It's important that people have the option to see them as discrepant data is sometimes is erroneously flagged as such. So the user needs to be able to see the data to unflag it when needed. I prefer making it a different color, like a light gray or something.
To implement this, the value could be moved to a different (discrepant) series list within the series map.
Select eps Aur from the 10 star list and retrieve from the database. There
is a data point from 2009 May 24 of 12.9 Visual that is plotted on the
chart despite being marked Discrepant. The plotting should probably ignore
such data points. Perhaps there could be an option to show them on the
chart if selected, but the default should be to hide them. Such data
points change the scale of the graph and distract from seeing the valid
data points plotted at a convenient scale.
Discrepant data points are now added to a "discrepant" series, not visible by default.
The scale arguably does need to be adjusted to match the min/max values in the dataset. Mike, Sara: do you concur? This means the scale could change upon *any* change in series visibility.
The color of all series is still random. That's next on the agenda.
Yes, I agree that the scale should reflect the current selection of series. For example, eps aur contains a discrepant at mag 12.9 (2454975.73889) and there is no point in crowding all the ~ mag 3 data points at the top of the graph. Also, if I start out with just V and Vis. and then add
opps.... that last comment was that if I add a new series that has data outside of the current range, the range should increase so that I notice those added points.
Magnitude range now scales dynamically with any change in the data to be plotted (e.g. series visibility). Mike, please evaluate the latest change and either close this if you are happy. Note that discrepant series and in fact all series color standardisation will be handled in another tracker entry.
The scaling fix was made in http://vstar.svn.sourceforge.net/viewvc/vstar?view=rev&revision=264
Looks and works great! This should now be marked closed. I can only close tracker IDs that I have started, so I don't have an option to close this one myself.
Try to get New Star from database and select Alpha Orionis from the 10 star pulldown menu. Then Change Series to include Discrepant. Not the scale of the Y axis. Then select Auto Range - Both Axes and note the scale change.
Then get alpha orionis again from the database. Do Auto Range - Both Axes to note the scale. Then Change Series to include Disrepant and note the scale. Do another Auto Range - Both Axes.
The scaling is not consistent with the selected Series, depending on what order you do the above actions.
Here is another example. Get Beta Lyrae from the database via 10 Stars. Drag the mouse to select a region to zoom in on. Then select Auto Range - Both Axes to zoom back out. The range for Y values looks too wide. Now Change Series to include Discrepants. VStar should probably show the narrower range when just viewing visual and the broader range only when Discrepant is part of the Series.
Mike, I'm thinking we should close this tracker entry with the discrepant series fix as described then open a new tracker entry with a new title and your two latest entries. What do you think?
I have opened a new tracker for the discrepant scaling issue at ID: 2905077
The original topic of this is fixed, and this one can now be closed.