[Andee Rubin]

I’m curious how CODAP decides to scale the Y-axis of a scatterplot, especially when a subset of points in a dataset are being shown.  I haven’t done an exhaustive exploration, but I’m surprised that sometimes the Y-axis rescales and sometimes it doesn’t – but I can’t tell you exactly the circumstances of either.  Is there something I should know that would help me predict when a Y-axis will rescale and when it won’t?

[Bill Finzer]

Hi Andee,

Interesting question. As the programmer for graphs I should be able to lay out the heuristics that determine whether a scatterplot (or any plot with a numeric axis) rescales.

The plot will rescale when:

• You press the Rescale button at the top of the graph’s “inspector.”
• An attribute is added to an axis or the legend.
• An attribute already on an axis or legend is changed to a different attribute.
• An attribute on an axis or legend is removed.
• One or more cases are added that have values for the plotted attributes that lie outside the current axis bounds.
• Points that have been hidden are shown and some of the values lie outside the current axis bounds.
• Cases that have been set aside are restored and some of the values lie outside the current axis bounds.

The plot will not rescale when:

• A case value is changed even if that value lies outside the current axis bounds.
• Cases are added but their values already lie within the current axis bounds.
• The graph is moved or resized.
• Points are hidden or their corresponding cases are deleted or set aside.

These heuristics evolved during the development of Fathom. I think you may be the first person expressing a desire to know what they are. And I don’t recall anyone ever asking that we change the rules, at least since the very early days.

 

Bill

[Author]

Posts