Develop Reference

View dataframe while debugging in VS Code

I'm trying to explore switching from PyCharm to VS Code. I can't find a way right now to view my pandas DataFrames in a tabular format while debugging.
When I right click on a df object, there is no option to view.
I have the python extension downloaded. Am I missing something?

Microsoft VSCode team finally made this feature available with latest update of the product. More details could be found in official blog
It works like a charm and is very intuitive. In short:
Set up a break point (by clicking at the left most point of code area, before line number)
Start debugging (Run menu at top have Start Debugging option)
When debugger stops at the debug point, find the required dataframe inside VARIABLES panel. (VARIABLES panel is inside Run and Debug area)
Right click on dataframe and select option View Value in Data Viewer. TADA :)

You can now print the DataFrame in the DEBUG CONSOLE:
From the Github issue mentioned in #Christina Zhou's answer.

My solution for viewing DataFrames in a tabular format while debugging is to simply copy and paste them into an Excel spreadsheet using
df.to_clipboard()
from the debug console. Even some of my colleagues running PyCharm are using this technique, since it gives you way more flexibility to inspect your data.

It seems like currently you can do it only using the Jupyter notebook in VS Code, using the variables explorer.

So it looks like this isn't a thing right now in VS Code.
If anyone wants to show their support for the development of this feature, I found this open issue here:
https://github.com/microsoft/vscode-python/issues/7063

you can use the view() function from xlwings library. It will show you the DataFrame in Excel:
import pandas as pd
from xlwings import view
df = pd.DataFrame({'A':[1,2], 'B':[3,4]})
view(df)
A better way would be to convert the function to pandas method:
from pandas.core.base import PandasObject
PandasObject.view = view
now you only need to type:
df.view()

Two more options for vscode are the following ones:
jupyter notebooks
saving to CSV and using the edit csv extension
Both require more effort but the view is more helpful.

The interactive shell looks like a good start. Right click the .py file in your explorer. You'll be able to view pandas dataframes from there.

Is it possible to use output_file instead of show_in_notebook when displaying LIME explanation results?

I am quite new to Python and I have just discovered LIME for model prediction interpretation, I have followed code from this tutorial: https://www.kaggle.com/emanceau/interpreting-machine-learning-lime-explainer
I am wondering if there is a way of displaying the explanations as they are shown in Jupyter notebook without using notebook, i.e running python on a text editor. I have seen something about output_file but I can't figure out how to implement it with lime. I am hoping to see something like what is shown in notebook:
show_in_notebook img
Is this possible, or do I need to start using Jupyter Notebook?

You don't really need to start a jupyter notebook to display the explanations as they are shown with the help of show_in_notebook function.
LIME package provides save_to_file function that allows one to save explanations to .html pages. In a python script it would be something like this:
explainer=lime.lime_tabular.LimeTabularExplainer(X.values,feature_names=X.columns)
exp = explainer.explain_instance(data_for_prediction, model.predict_proba)
exp.save_to_file('lime.html')
There's also as_pyplot_figure that creates a barchart explaining the prediction. The output can be saved with matplotlib.savefig.

Semi-Interactive Pandas Dataframe in a GUI

There are a number of excellent answers to this question GUIs for displaying dataframes, but what I'm looking to do is a bit more advanced.
I'd like to display a dataframe, but have a couple of the columns be interactive where the user can manually overwrite values (and the rest be static). It would be useful to have "total" rows that change with the overwritten values and eventually have some interactive buttons around the dataframe for loading and clearing data.
QTPandas looks promising, but appears to be dead as it is build off of a really old version of Pandas (0.17.1). Can this be done in QT? Is something else better?

I love Rstudio as my IDE as I can not only view all objects created but I can also edit data in the IDE itself. There are many other great features too.
And you can use R Studio for Python coding too (using reticulate package).
Spyder too gives this feature of viewing or editing the data frame.
However, if you're looking for a dedicated GUI with drag & drop features, you can use Pandas GUI.
Features of pandasgui are:
View DataFrames and Series (with MultiIndex support)
Interactive plotting
Filtering
Statistical summary
Data editing and copy / paste
Import CSV files with drag & drop Search toolbar
It's first version was released in Mar 2019 & still developing. As of date, you can't use it in Colab

While not a GUI in itself, XLWings leveraged Excel as a GUI and makes pandas dataframes interactive for users and was our library of choice.

Is it possible to redirect cell output in jupyter

I am using jupyter and jupyter-nbconvert to create a html presentation. However, I have some cells that produce an output image that I want to share on a separate slide. Is it possible to redirect the output of one cell to its own slide?

You might want to consider using Damian Vila's Jupyter extension RISE. It provides some of the control you need for how cells are displayed in slides.
It is flagged by the latest Jupyter (3.6) as possibly not compatible, but I've seen no problems using it so far.

Python pandas display dataframe as a table with tabs and not spaces

How can I print or display a pandas.DataFrame object like a "real" table? I mean using only one tab between columns and not more spaces. In IPython Jupyter Notebook I can use the following code to get a "real" table style:
from IPython.core.display import display
display(df.head(50))
instead of print(df.head(50)) which uses spaces.
Is there any same in IPython console using Spyder? I did no find a proper pd.set_option() value...

Unfortunately the Jupyter Notebook is a specialized environment, which knows how to tell objects to render themselves as HTML inside the Jupyter (web) interface. I don't believe Spyder has this capability.