Description of need: One use of the observation calendar is to look back at days where you posted many observations, or to perhaps re-ID some of your observations from a specific time. The dates on the calendar are color-coded by the amount of observations were recorded on that day (i.e. a day with 1 observation may appear a very light gray, whereas a day with 20 would appear slightly darker). This can be often be helpful, but it would be much more helpful if days with 50 observations have a different color than days with 100. Or even 400 observations, for some users. At the moment, the color doesn’t appear to change above 50ish.
Feature request details:
I’m proposing to change the way dates are color-coded based off of the observation count. Perhaps continuing to make the days darker after 50 observations, or lightening days before around 100ish. Currently, I think the difference between 10-30 observations is exaggerated, but the difference between 40 and 100+ observations is barely anything.
how do you pick the right arbitrary cutoff that works for the most people? why is 100 better than 50 for most people?
if you wanted the scaling to vary by user, it seems to me like you would just scale from 0 to whatever the max daily observation count is for a particular user.
Good point. In my second proposal, though, everything below 50 observations remains the same as it does currently. This would mean the calendar would look the same for users who don’t typically post more than 50 observations a day. So it’s more like removing an arbitrary cut-
honestly, i think it should be the lowest of low priorities. my take is that a separate calendar page for a given user really shouldn’t exist and should eventually be removed. in its place, whenever the Explore Page is finally revamped, it should include a temporal view tab where you could select from either a time series bar chart / line graph or a calendar heatmap visualization for the observations retrieved. this would allow folks to get a better understanding of iNat data temporally, and it could be used for any set of data that could be retrieved from the Explore Page (not limited to a particular user).
the experimental comparison tool already is sort of structured this way, with a species tab, a map tab, and a history (line graph) tab.
are you arguing that something that can handle only one specific use case shouldn’t eventually be replaced by something that would be much more flexible, handling that original use case plus many other cases?
I think this idea would provide the most intuitive way for shading the different days on the calendar. If say, your max observations in one day ever was 58, that day would be black, and all other days would be put on an according scale, relative to 58. Good thinking.
I want to keep the personal calendar. It’s a nice way to visualise your progress and all that. I don’t really understand your idea for what will be in its place, though.
I do agree that this should probably be the least of iNaturalist’s concerns as of now.
I wonder if perhaps instead of selecting arbitrary cut offs they could scale automatically and provide a key. This would allow it to be useful regardless of how common an organism is.
Edit: Adding more details before people reply. Having fleshed it out in my head more.
I am not a statistician, but to automatically break things up I can think of two ways off hand.
You could just break them into even segments based off taking the highest number of observations, rounding it upwards to a nice number, and dividing by how many marker colors you have. This gives a nice even distribution of markers and I think would work well in most cases.
But let’s say you have an organism whose observations, for whatever reason, do not fit nicely into smooth even categories. For that we could break it into uneven segments using something like Jenk’s natural breaks algorithm, which would try to cluster the number of observations per day into groups that are similar in number.
Using Jenks might be overcomplicating it some though, I couldn’t actually think of a situation where even segmentation doesn’t work, except to break outliers out into their own bins, or if observations of an organism pick up suddenly and rapidly in the middle of a month. Maybe on a yearly calendar?
Anyhow, done with the edit, just wanted to flesh this out some before anyone replied.
