ENH: Easy overview of content of dataframe

[thomasfrederikhoeck]

Before building any models and making any visualsations it is useful to inspect the data set first.
A function that gives a quick overview of the content could be the following. It is build for Jupyter Notebooks:

import pandas as pd
from IPython.display import display


def display_all(df):
    df_disp = pd.DataFrame(index=df.columns, columns=['type', 'count', 'unique', 'NaN',
                                                      'mean', 'median', 'std', 'min', 'max',
                                                      'values'])
    i = 0
    for col in df.columns:
        type_i = str(df[col].dtype)
        NaN_i = np.sum((pd.isnull(df[col])))
        count_i = len(df[col])-NaN_i
        unique_i = len(df[col].unique())
        col_unique = df[col].unique()
        if len(col_unique) > 1:
            val_i = str(col_unique[0]) + ', ' + str(col_unique[-1])
        else:
            val_i = str(col_unique[0])
        if type_i == 'object':
            mean_i = np.NaN
            median_i = np.NaN
            std_i = np.NaN
            min_i = np.NaN
            max_i = np.NaN
        else:
            mean_i = np.nanmean(df[col])
            median_i = np.nanmedian(df[col])
            std_i = np.nanstd(df[col])
            min_i = np.nanmin(df[col])
            max_i = np.nanmax(df[col])
        df_disp.iloc[i, :] = [type_i, count_i, unique_i, NaN_i, mean_i, median_i, std_i, min_i,
                              max_i, val_i]
        i += 1
    with pd.option_context('display.max_rows', None, 'display.max_columns', 10):
        display(df_disp)