Interactive plots for exploring and selecting points from DataFrames in a Jupyter notebook. Simple as passing DataFrame df to a dfbrowser instance with
from dfbrowser import dfbrowser
browse = dfbrowser(df)This will produce an interactive scatterplot of numeric column data from your DataFrame. Here's an example screenshot with random data:
You can pick which columns to plot from the dropdown menus. Click near points to select them. The pandas Series associated with the selected row is passed to user defined function funct (either defined upon initialization or by setting the dfbrowser.funct parameter). This function must be able to accept a pandas Series object as input. The standard output for print calls and exceptions will be directed to an output widget below the figure (this can be cleared with a call to dfbrowser.clearoutput()). You can also access the currently selected index and row as dfbrowser.selectedindex and dfbrowser.selectedrow.
You must use the ipympl >v0.5.4 backend with the notebook magic command %matplotlib widget, and enable these two Jupyter notebook extensions from the Terminal:
jupyter nbextension enable --py --sys-prefix widgetsnbextension
jupyter nbextension enable --py --sys-prefix ipympl
I use this tool to explore a large number of astronomical light curves for many stars, each of which has a number of numerical values associated with it (e.g., position on sky, average brightness, dominant timescale of variablility). I use dfbrowser to see how these stars are distributed in different parameter spaces, and then I connect a function that displays the associated light curve (loaded from a column containing filenames usually) for any point I click on.
Raise an issue if you would like to see any other functionality. I plan to let marker sizes and colors vary with third and fourth columns eventually.