Implement plotObservedVsSimulated()

[IndrajeetPatil]

TODO

• Implement interpolation function
• Implement plotObservedVsSimulated()
• tests for plotObservedVsSimulated()
• check comments
• handle empty DataCombined objects
• Address Pavel's code review comments (Implement plotObservedVsSimulated() #945 (comment))

Features that need to be implemented in tlf, and thus should not block this PR from being merged:

• Supporting fold distance in linear scale TLF-Library#332
• Add caps to error bars TLF-Library#333

Related to #674

Discussed in #809

Closes #804
Closes #924
Closes #1010
Closes #1011
Closes #1012

[IndrajeetPatil]

@msevestre, @Yuri05 This PR is ready for a review. Thanks.

[IndrajeetPatil]

Observed data is plotted with error bars now.

Note that the error bar is missing in the plot below for some data points because the error value is 0.

[Yuri05]

Should we display the error bars in the same way as usually?

[IndrajeetPatil]

Should we display the error bars in the same way as usually?

This would be easy to do by setting height to some non-zero value in geom_errorbarh().

I personally find those vertical edges distracting. Plus, they don't convey any additional information, so it's just adding clutter to an already busy plot. So I'd prefer not adding the edges.

I'm curious to hear what others think.

[Yuri05]

they don't convey any additional information, so it's just adding clutter to an already busy plot. So I'd prefer not adding the edges.

In general, I agree.
However we display error bars always like this (example) and looking at the plots by others - same in the most cases (not always though)

[IndrajeetPatil]

@Yuri05 Does this mean that we should also be displaying error bars with caps in profile plots?

Currently, we have error bars without the caps:

[PavelBal]

Does this mean that we should also be displaying error bars with caps in profile plots?

Hmm yes, I did not realize this but yes, it is usual to display caps for error bars.

[Yuri05]

Does this mean that we should also be displaying error bars with caps in profile plots?

I would say yes :)

[IndrajeetPatil]

Okay, I've created an issue it tlf to add caps to error bars: Open-Systems-Pharmacology/TLF-Library#333

Once it is added there, it will automatically be included in the ospsuite plots as well.

[IndrajeetPatil]

@PavelBal and @Yuri05 I've created issues for features that need to be implemented in tlf, and thus should not block this PR from being merged:

• Supporting fold distance in linear scale TLF-Library#332
• Add caps to error bars TLF-Library#333

I don't want to keep this pending anymore, since this is blocking me from working on #1002.

[Yuri05]

@PavelBal and @Yuri05 I've created issues for features that need to be implemented in tlf, and thus should not block this PR from being merged:

* [Supporting fold distance in linear scale TLF-Library#332](https://github.com/Open-Systems-Pharmacology/TLF-Library/issues/332)

* [Add caps to error bars TLF-Library#333](https://github.com/Open-Systems-Pharmacology/TLF-Library/issues/333)

I don't want to keep this pending anymore, since this is blocking me from working on #1002.

fine for me

[IndrajeetPatil]

Thanks, Yuri.

@PavelBal Let me know if you have any other comments over and above the ones that have already been addressed.