change graph style for percentile — thicker median

[cypressf]

I might consider using geom_ribbon to make a filled-in range of color for the percentile ranges.

[holak]

@cypressf that would be great. The figures in the Outlook Report look something like this: With the thick line representing the median, the dashed lines the interquartile range (25th-75th percentiles) and the shaded area the 5th-95th percentiles)

[cypressf]

How's this look?

[holak]

Awesome!!

[cypressf]

Is it mathematically sound to say that the values in the 25th to 75th percentile have a 50% likelihood and the values from the 5th to 95th percentile have a 90% likelihood? I wanted to add a legend like so:

[cypressf]

Maybe "Confidence Interval" is the term I'm looking for, not "Likelihood"

[cypressf]

I'll close this, and if confidence interval is not the correct term, lmk and I can change the legend title.

[holak]

@cypressf "Confidence Interval" works, but is it too technical for general audiences? Other potential options: "Probability Range", "Probability Bounds", even "Likelihood" which less technically precise could work. Let's ask Sergey and Adam for their thoughts as well. Also would it be too long for the labels to be: "90% (5th-95th percentile)", "50% (25th-75th percentile)"? Maybe that's too much and not necessary? Finally, is there a way to also have a label for the black line as the median? I guess that gets tricky because median wouldn't fall under the same label heading as the probability bounds... maybe a reason to go with the more general "Likelihood"? Let's give it some more thought and circle back. I really like the shaded areas though.

[holak]

@cypressf the consensus was to change the legend title from "Confidence Interval" to "Likelihood Range"

[cypressf]

Cool. I'll rename it!