fagan2888 / dataframe_visualiser

Make simple high-level visualisations of a Pandas dataframe to help when starting new data projects

dataframe_visualiser

Make simple high-level visualisations of a Pandas dataframe to help when starting new data projects with Seaborn:

• Summarise single variables as a grid of plots
• Summarise each variable and its dependency on a boolean column as a grid of plots
• Show high-level overview of a subsampled set of cells as a matrix

This is a Python 3 project (and with some tiny edits it would run in Python 2.7).

To run:

>>> # assuming IPython and Python 3.4
>>> import visualise_dataframe
>>> visualise_dataframe.summarise(df)  # draw single-variable plot
>>> visualise_dataframe.summarise(df, dependent_col='some_col_name_in_df')  # draw dependent-variable plot
>>> visualise_dataframe.show_cells(df)  # high-level overview of cells as a matrix

If you run the script then the example will load the Kaggle Titanic competition's train.csv file and will draw a single-variable plot.

Examples

Using the Kaggle Titanic example http://www.kaggle.com/c/titanic-gettingStarted data set we can draw each variable independently:

We can also ask it to draw each variable when dependent on a boolean (in the case against Survived):

Finally with show_cells we can subsample a set of rows and show their dtypes and whether they're NaN or not, to get an idea about the relationship between NaNs and other data:

TODO

• don't subsample df and then analyse, instead analyse before subsampling to count NaNs in full columns
• add legend to show_cells to identify the colours!
• consider analysing text cells to show e.g. length of each cell, two tone for binary categories etc for show_cells