Open chriddyp opened 7 years ago
https://codepen.io/etpinard/pen/gRzGXg
Do 1D log-axes histogram even work? Is that even a thing?
Here's the perspective of the use who reported it:
This one has some interesting theoretical implications: It's not just a matter of displaying the histogram on a log axis (showing a heatmap or bar trace on a log axis works fine). You also need to space the bins logarithmically, and if there's normalization it amounts to something like "counts per decade" instead of whatever units are on the axis.
Also this would give us more motivation to include the bounds of the bin in hover labels, not just the center - see #69. Or, even better, the bounds of the data in the bin... for example if your bin is 1.5 - 3.5 but your data is all integers, it should say 2 - 3.
https://codepen.io/etpinard/pen/gRzGXg
Do 1D log-axes histogram even work? Is that even a thing?
totally a thing, and would love to use plotly for this. Is this in the works...?
totally a thing,
Could you share an example/screenshot from another library to help us out?
currently using numpy to set bins and plotly bar as work around
log option in matplotlib sets y axis to log scale rather than x. No great option in python that I'm aware of, other than setting bins manually.
Log scale x axis is desirable in the case of highly skewed distribution. In my own experience is this is common with geoscience data where GSLIB Histplt is often used
From my point of view, the 1D histogram should show like a normal histogram of the logarithm of the data, and actually my current workaround is to do that and then manually set the non logarithm values for the ticks. The only problem is that this only works well with static bins. Example:
I, too, would like to use log axes in a histogram2d. Currently, it is just odd. For instance, using the "2D Histogram of a Bivariate Normal Distribution with Plotly JS" example and just changing the yaxis to type: 'log', you get a strange row of data: https://codepen.io/manthey/pen/VNjZMy If you manually change the range of the y axis, it looks like all of the data is offset, but otherwise somehow binned: https://codepen.io/manthey/pen/EJyYLW
+1 for this feature
having the same issue here, is there a workaround?
+1
+1
+1
This still seems like a good idea! FYI to folks coming across this issue: If your organization needs this feature, then reach out to our services team to fund the development: https://plotly.com/products/consulting-and-oem. This is how many new features are supported & developed 🍻
This issue has been tagged with NEEDS SPON$OR
A community PR for this feature would certainly be welcome, but our experience is deeper features like this are difficult to complete without the Plotly maintainers leading the effort.
Sponsorship range: $10k-$15k
What Sponsorship includes:
Please include the link to this issue when contacting us to discuss.
Bumping... no reason this should still be open. I can't do log_z=True
still...
@gvwilson I see some recent activity on this. Does it mean it will be soon available ? Would be an absolute game changer for my case. Many thanks
hi @paulduf - I started as the Engineering Manager for the Libraries team 14 weeks ago and am still sorting through backlog. No idea when/if someone will be able to get to this one, but I'll post updates here. cheers - @gvwilson
as reported in the community forum: https://community.plot.ly/t/2d-histogram-with-log-axes-supported/4711
example: https://plot.ly/~chris/18409/