quickplotter 0.2

Instantly generate common EDA plots without cleaning your DataFrame

Instant EDA (Work in progress!)

Instantly generate common exploratory data plots without having to worry about cleaning your data.

The code is hosted on PyPi, the Python Package Index here.

It can be installed by running

pip install quickplotter==0.1

To setup the proper development environment, run conda env create -f environment.yml

Usage:

plotter = quickplotter.QuickPlotter(df: pd.DataFrame) #creates a QuickPlotter object with the given DataFrame

plotter.common(subset=['correlation', 'percent_nan']) #plots correlation between features, and percent nan in each column

plotter.distribution(column_subset=df.columns[0:4]) #plots distributions for the first four columns in the DataFrame

The quickplot module works mainly with two specifications, subset and diff.

For any subset-like list, the items in the list will be used. For any diff-like list, all items except those in the list will be used.

To specifiy column subset's or diff's, call each plot individually or call .common with the column_subset or column_diff attributes (need to be added as of 6/18/20).

Ideas so far:

Number of NaN's in each column (done)

Percent of Nan's in each column (done)

Correlation matrix (done)

distribution matrix for all features (done)

univariate distribution of each feature (bar + kde for numeric, just bar for categorical)

time series distribution of numeric features if we can infer a timestamp column (look at: this)

pairplot of all numerical everything if number of columns is manageable (done)