Edit- Closing this thread since the number of feature requests from users other than myself has grown pretty large so going forward let's track each one in a new issue. If you have any feature request that doesn't have an open issue already please create one

I'll be working on the list below still but not adding to it anymore. if you see something listed that you really want implemented or needs more discussion feel free to open an issue if it doesn't have one.

Planned Enhancements

• Set up CICD and better unit testing
• Upgrade from PyQt5 to PyQt6
• Add image HTML export for Grapher (Plotly)
• Add checkbox to export dialog for whether to include index. Leave drag & drop export assuming index=False
• Add progress indicator and status bar & log for all messages and actions
• Column reordering
• Column hiding/deleting
• Improve Statistics tab (make columns selectable to show more details such as...)
  • unique values counts
  • histogram
  • Count and % missing (NaN)
• Change column data types
• Put new Grapher plots in tabs instead of overwriting previous plot
• Data display format (float & date)
• Allow multi-selection and rearranging of DataFrames in left nav menu
• Rename dataframes
• Drop NA / fill NA
• Save workspace state (so you can close PandasGUI and re-open it later with all the same data/plots/filters)
• Outlier detection
• JMP-style 'calculated' columns which are columns with values defined by a function, that get automatically recalculated after any change to the DataFrame
• Lasso selection in graphing UI that can filter selected points from the DataFrame
• When you click on a cell and then click on a row, it should highlight the row (as when you just click on a row)

Done

• Find search bar
• Grapher tab (drag & drop columns to do plotting)
• Column sorting
• Filters tab
• Make tabs detachable and dockable
• Edit DataFrame by double clicking a single cell and typing a new value
• Sort DataFrame by a column by left clicking the header
• Add DataFrame graphing tab
• Filters that can be toggled on and off
• Action history log
• Custom kwargs in Grapher
• Import / Export CSV
• Copy/Paste support in DataFrame viewer
• Add Reshaper (pivot / melt)
• Automatic date parsing on import
• Dark mode
• Generate runnable code from change history to replicate steps (including plotting)
• Add right-click context menu on column headers
• Add nunique column in the Grapher drag and drop interface
• Option to delete DataFrame

Will not add

• Automatically sync changes between DataFrames in GUI and iPython
• Support for remote servers like Google Colab or Jupyterhub. This is not possible, you can only render PyQt5 widgets on the same machine it is running on.

Make group by on lines optional

Just tracked an issue where pretty much any time series with more than a handful of groupings just hangs pandasgui.

This is because group by explodes with each grouping, getting lots of zero values (or even null values if data has holes), creating a sparse dataframe in the billions of rows... The longer the time series or unique values in grouping, the worse it gets. ie. 30K values in X, col A with 14 values values and col B with 5 unique values, That's over 2B rows, starting with a dataframe of only 30K rows... Typical laptops start to really struggle at that point, and plotly just not displaying anything in many cases.

Ask

• Maybe disable that until we have derived values?
• or add a checkbox to enable this?
• Or perhaps only include X and color. (Even then, if color is a continuous scale instead of a categorical, this could once more explode...)

Side benefit is that data is not masked when disabled.

When trying to plot a dataframe on Mac (python 3.8, latest pandasgui from this repository), I am getting an error saying webgl is not available. I checked all browsers I have and they all support webgl (meaning when I go to the webgl site, I can see the graph - it is possible this is webgl2 not 1).

The dataset I used is here: https://drive.google.com/file/d/1qlUt1_X_0BxbmFDP5YgAqIQBp3osP4jz/view?usp=sharing

Hi just upgraded to 0.2.8 now and then I starting to get this error

TypeError                                 Traceback (most recent call last)
<ipython-input-4-7fe66a702bc1> in <module>
      1 from pandasgui import show
----> 2 show(df)

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandasgui\gui.py in show(settings, *args, **kwargs)
    362     kwargs = {**kwargs, **dataframes}
    363 
--> 364     pandas_gui = PandasGui(settings=settings, **kwargs)
    365     pandas_gui.caller_stack = inspect.currentframe().f_back
    366 

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandasgui\gui.py in __init__(self, settings, **kwargs)
     74         # Adds DataFrames listed in kwargs to data store.
     75         for df_name, df in kwargs.items():
---> 76             self.store.add_dataframe(df, df_name)
     77 
     78         # Default to first item

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandasgui\store.py in add_dataframe(self, pgdf, name)
    413         pgdf.store = self
    414 
--> 415         pgdf.df = clean_dataframe(pgdf.df, name)
    416 
    417         # Add it to store and create widgets

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandasgui\utility.py in clean_dataframe(df, name)
    264     converted_names = []
    265     dtypes_old = df.dtypes
--> 266     df = parse_dates(df)
    267     dtypes_new = df.dtypes
    268     for col_name in [df.columns[ix] for ix in range(len(dtypes_new)) if dtypes_old[ix] != dtypes_new[ix]]:

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandasgui\utility.py in parse_dates(df)
    229 def parse_dates(df: Union[pd.DataFrame, pd.Series]):
    230     if type(df) == pd.DataFrame:
--> 231         return df.apply(
    232             lambda col: pd.to_datetime(col, errors='ignore') if col.dtypes == object else col,
    233             axis=0)

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandas\core\frame.py in apply(self, func, axis, raw, result_type, args, **kwds)
   7763             kwds=kwds,
   7764         )
-> 7765         return op.get_result()
   7766 
   7767     def applymap(self, func, na_action: Optional[str] = None) -> DataFrame:

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandas\core\apply.py in get_result(self)
    183             return self.apply_raw()
    184 
--> 185         return self.apply_standard()
    186 
    187     def apply_empty_result(self):

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandas\core\apply.py in apply_standard(self)
    274 
    275     def apply_standard(self):
--> 276         results, res_index = self.apply_series_generator()
    277 
    278         # wrap results

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandas\core\apply.py in apply_series_generator(self)
    288             for i, v in enumerate(series_gen):
    289                 # ignore SettingWithCopy here in case the user mutates
--> 290                 results[i] = self.f(v)
    291                 if isinstance(results[i], ABCSeries):
    292                     # If we have a view on v, we need to make a copy because

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandasgui\utility.py in <lambda>(col)
    230     if type(df) == pd.DataFrame:
    231         return df.apply(
--> 232             lambda col: pd.to_datetime(col, errors='ignore') if col.dtypes == object else col,
    233             axis=0)
    234     elif type(df) == pd.Series:

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandas\core\tools\datetimes.py in to_datetime(arg, errors, dayfirst, yearfirst, utc, format, exact, unit, infer_datetime_format, origin, cache)
    799                 result = result.tz_localize(tz)
    800     elif isinstance(arg, ABCSeries):
--> 801         cache_array = _maybe_cache(arg, format, cache, convert_listlike)
    802         if not cache_array.empty:
    803             result = arg.map(cache_array)

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandas\core\tools\datetimes.py in _maybe_cache(arg, format, cache, convert_listlike)
    171     if cache:
    172         # Perform a quicker unique check
--> 173         if not should_cache(arg):
    174             return cache_array
    175 

~\AppData\Local\Programs\Python\Python39\lib\site-packages\pandas\core\tools\datetimes.py in should_cache(arg, unique_share, check_count)
    135     assert 0 < unique_share < 1, "unique_share must be in next bounds: (0; 1)"
    136 
--> 137     unique_elements = set(islice(arg, check_count))
    138     if len(unique_elements) > check_count * unique_share:
    139         do_caching = False

TypeError: unhashable type: 'list'

Environment OS: (eg. Windows 10) Python: (eg. 3.9)

Hi, thanks for your awesome work. I love it

I didn't use every features of pandasgui yet. but i'll try. And maybe I expect myself to write very basic tutorial for my students.

but there's few suggestion for this project.

1. it's good to see this in colab.

2. when pandasgui didn't work in colab. i tried to use it on jupyter lab. (and it worked !) however as my local pc has dark theme. it didn't work properly.

Here are my environments

• macOS catalina
• python version 3.8.3
• pandasgui version 0.2.7
• pandas version 1.1.2

thanks again.

I noticed that you're looking for feedback... and I think this project is on to something...

My dream DataFrame viewer would work like this:

from pandasgui import DataFrameViewer
viewer = DataFrameViewer()
viewer.open()  # opens a GUI window
viewer.bind(df1)  # tab 1 in the viewer
viewer.bind(df2)  # tab 2 in the viewer
viewer.bind(df3)  # tab 3 in the viewer

Now when you change a cell value in the GUI, the application "pushes" the code to the python/ipython shell i.e., you see df1.at['row', 'col'] = 10 in the shell.

Moreover, if you type df1.at['row', 'col'] = 20 in the shell, the value automatically changes in the GUI. I think such an application would be 1000+ Github stars in no time.

Hi @adamerose ,

I am trying pandasgui in some python script, unlike in interactive cmd, it shutdown immediately after running. I find settings={'block': True} is necessary, would you please figure it out in the demo examples? Thank you and pandasgui is nice tool!

Purpose

Adding autocompletion to the filter QLineEdit widgets. This is to get feedback. I would think this is better than copy / paste or drag and drop, maybe? If this works ok, I'd like to expand and do proper filtering of valid values based on column in a future PR.

This doesn't really close #88 as I think the column header filtering is probably what a lot of people really want, but that is too much of time budget for me ATM.

Added:

• constant for max cardinality of categoricals to be included in autocompletion (should be a user setting...)

• checkbox below the filter box to enable/disable autocompletion

• QCompleter based on a list of columns (backtick delimited) and values for low cardinality categoricals (quote delimited)

• case insensitive suggestions

• start with backtick to get list of columns only

• type the numexpr as usual, ie ==, in (), !=, then to select an available value, use quote (")

Beside doing feature requests or creating bugs, there's a good bit of stuff that would benefit from a mailing list to discuss some development issues, UI/UX feedback from users etc.

PLEASE FILL OUT THE TEMPLATE

Hello, thank you for this amazing app, it is really helpful.

Describe the bug

Some plots works and other don't. Histogram is fine, as well as Contour and Heatmap.

But the two main ones that I want to use (scatter, line) don't work and shows a message in the GUI when I click on the "Finish" button :

" WebGL is not supported by your browser - visit https://get.webgl.org for more info"

I think (not 100% sure) that i remember these types of plots were working like a week or two ago, but nothing really changed in my code since (no pip packages upgraded)

The python console says WebEngineContext used before QtWebEngine::initialize() or OpenGL context creation failed.

each first time I use the function show() from pandasgui. I think this message is related to the issue but I have no idea how to fix it.

Environment OS: Mint 20 cinnamon Python: 3.8.5 conda virtual envrionment IDE: VsCode python interactive console by copy/pasting the code, OR by running the python script, same issue

Package versions

console.log("I'm a code block!");

Any help would be much appreciated. I did read this related issue but didn't find any fix for my case there https://github.com/adamerose/pandasgui/issues/25

Thank you!

Hi there,

I wonder if I can run pandasgui in google colab. I tried to use the following for a virtual display but did not work

!pip install pyvirtualdisplay from pyvirtualdisplay import Display Display(visible=0, size=(1400, 900)).start()

Any suggestion is greatly appreciated.

Cheers, Hossein