Answer by Niko Fohr for Find out max value from a python dict

Pandas is very effective tool for handling tabular data like this. You could create a pandas DataFrame from the data:

import pandas as pddf = pd.DataFrame(d).Tdf.columns = ('group', 'place', 'value')

and then just print out the maximum values

df[df['value'] == df.groupby('group')['value'].transform('max')]

which gives

Out[41]:   group                                    place     value1      1                                  Toronto       0.810     2                                 Arkansas  0.88714114     3                                  Toronto       0.823     4          Nashville International Airport       0.827     4                                   Turkey       0.830     5                         Hebron, Kentucky       0.838     5                   County (United States)       0.842     6                                  Ontario       0.848     7                                   London  0.95458558     8                 UNITED STATES OF AMERICA  0.84546361     9      RaleighDurham International Airport       0.865     9                                    Piauí       0.872    10                               New Jersey       0.881    10  Morrisville, Bucks County, Pennsylvania       0.884    10                            United States       0.888    10                New Brunswick, New Jersey       0.8

If you want to get the output in the original format, you can use df.to_dict

In [47]: df[df['value'] == df.groupby('group')['value'].transform('max')].T.to_dict(orient='list')Out[47]:{1: ['1', 'Toronto', 0.8], 10: ['2', 'Arkansas', 0.8871413469314575], 14: ['3', 'Toronto', 0.8], 23: ['4', 'Nashville International Airport', 0.8], 27: ['4', 'Turkey', 0.8], 30: ['5', 'Hebron, Kentucky', 0.8], 38: ['5', 'County (United States)', 0.8], 42: ['6', 'Ontario', 0.8], 48: ['7', 'London', 0.9545853137969971], 58: ['8', 'UNITED STATES OF AMERICA', 0.8454625606536865], 61: ['9', 'RaleighDurham International Airport', 0.8], 65: ['9', 'Piauí', 0.8], 72: ['10', 'New Jersey', 0.8], 81: ['10', 'Morrisville, Bucks County, Pennsylvania', 0.8], 84: ['10', 'United States', 0.8], 88: ['10', 'New Brunswick, New Jersey', 0.8]}

Short explanation

• Pandas dataframes can be created using dictionaries as arguments. The values should be lists. The .T takes just transpose of the table.
• The df.groupby('group')['value'] returns a SeriesGroupBy object, which behaves very much like a regular pandas.Series object. With that we can calculate the maximum value for each group, using the transform method.
• The df['value'] == df.groupby('group')['value'].transform('max') creates a boolean mask for selecting the maximum rows by df[mask].