Allow users to hide 'uid' and 'from_uid' from the table

[kamil-kaczmarek]

Hi @danthe3rd,

Thanks for the nice library. Very lightweight, very useful 🙂

Background
When I have a lot of axes on the parallel plot I always drop uid and from_uid axes, because they add some unnecessary clutter to the chart. I still want to keep these two axes in the table.
In my case I have other axes than encode runs.

Request
Add an option (maybe to the display function) to not display uid and from_uid axes on the chart.

Best,
Kamil

[danthe3rd]

Hi,

There is a hide setting you can use to hide columns in the parallel plot. You can have a look here for an example: https://facebookresearch.github.io/hiplot/experiment_settings.html#frontend-rendering-settings

I hope this helps :)

[kamil-kaczmarek]

Hey @danthe3rd,

Yes, it solved my problem, thank you.

[bensdm]

Is it possible to hide on parallel plot but display on the table below?

[danthe3rd]

Is it possible to hide on parallel plot but display on the table below?

All the columns are always displayed on the table. The hide setting only refers to the parallel plot part.

[bensdm]

indeed sorry i made a typo in my code, but is is possible to hide from uid in the columns?

[danthe3rd]

Not yet no. Maybe in a future version :)

[haw230]

Just wanted to add I would like this change too! It would make the table a lot cleaner to be able to drop uid and from_uid.

[danthe3rd]

Reopening to track it

[danthe3rd]

I've just pushed 0.1.19 (you can upgrade with pip install -U hiplot), which adds support for hiding columns in the table.

The documentation contains an example: https://facebookresearch.github.io/hiplot/experiment_settings.html#frontend-rendering-settings. You can now add this to hide the uid and from_uid columns:

exp.display_data(hip.Displays.TABLE).update({
    # Don't display `uid` and `from_uid` columns to the user
    'hide': ['uid', 'from_uid'],
})

[kamil-kaczmarek]

Great to see this change. Thanks @danthe3rd

[AndyJZhao]

Is it possible to hide the "uid" and "from_uid" in the jupyter_notebook? I'm getting the "uid" columns occasionally while plotting with hip.Experiment.from_dataframe(df).display(). See the right-most 'uid' column in the pic below.

[GoldenCorgi]

@Andy-Border

Is it possible to hide the "uid" and "from_uid" in the jupyter_notebook? I'm getting the "uid" columns occasionally while plotting with hip.Experiment.from_dataframe(df).display(). See the right-most 'uid' column in the pic below.

Does this code below work for you?

exp = hip.Experiment.from_dataframe(df)
exp.display_data(hip.Displays.PARALLEL_PLOT).update({'hide': ['uid']}) # Change ['uid'] to a list of column names to be hidden. E.g. ['uid', 'dropout']
exp.display()

Reference: https://colab.research.google.com/github/facebookresearch/hiplot/blob/main/examples/HiPlotColabExample.ipynb

Should work in Colab and other Jupyter environments

If it doesn't work, do open a new issue, I'll work on it if it's a bug!

[AndyJZhao]

@Andy-Border

Is it possible to hide the "uid" and "from_uid" in the jupyter_notebook? I'm getting the "uid" columns occasionally while plotting with hip.Experiment.from_dataframe(df).display(). See the right-most 'uid' column in the pic below.

Does this code below work for you?

exp = hip.Experiment.from_dataframe(df)
exp.display_data(hip.Displays.PARALLEL_PLOT).update({'hide': ['uid']}) # Change ['uid'] to a list of column names to be hidden. E.g. ['uid', 'dropout']
exp.display()

Reference: https://colab.research.google.com/github/facebookresearch/hiplot/blob/main/examples/HiPlotColabExample.ipynb

Should work in Colab and other Jupyter environments

If it doesn't work, do open a new issue, I'll work on it if it's a bug!

Hi GoldenCorgi,
Thanks for your prompt reply and valuable code snippet. It worked🥂!
Yet it would be very helpful to drop the "uid" and "from_uid" columns in the table below the plot, Q1. is there any method to achieve this?
Q2. By default, the table is inconveniently sorted by the "uid", which is sorted alphabetically i.e. 1,10,100, etc. (check the pic below), is there any method to pass a sorted_by argument to the table? (instead of manually click on the column title)🧐

[danthe3rd]

Hi :)

Yes you can do both. The documentation is here

I didn't test it, but this code should work:

exp = hip.Experiment.from_dataframe(df)
exp.display_data(hip.Displays.PARALLEL_PLOT).update({'hide': ['uid']}) # Change ['uid'] to a list of column names to be hidden. E.g. ['uid', 'dropout']
exp.display_data(hip.Displays.TABLE).update({'hide': ['uid']}) # Same here
exp.display_data(hip.Displays.TABLE).update({'order_by': [('lr', 'asc')]}) # Order by LR asc
exp.display()

[AndyJZhao]

Hi :)

Yes you can do both. The documentation is here

I didn't test it, but this code should work:

exp = hip.Experiment.from_dataframe(df)
exp.display_data(hip.Displays.PARALLEL_PLOT).update({'hide': ['uid']}) # Change ['uid'] to a list of column names to be hidden. E.g. ['uid', 'dropout']
exp.display_data(hip.Displays.TABLE).update({'hide': ['uid']}) # Same here
exp.display_data(hip.Displays.TABLE).update({'order_by': [('lr', 'asc')]}) # Order by LR asc
exp.display()

Thank you mate, this works like a charm 🤗 !