Tools for calculating dominance hiearchies in Python. Working on including:

• Elo Scores Albers & de Vries 2001
• David's Scores David 1987
• ADAGIO Douglas et al 2018
• Randomized Elo Sánchez‐Tójar et al 2017
• Linearity Statistics

First, import packages:

import pandas as pd
import hierarpy as hp

Load an example dataframe (provided in hierarpy/test_dfs)

df = pd.read_csv('test_dfs/df1.csv')

This dataframe looks like the following:

We can tabulate this dataframe using hierarpy's matrix_from_dataframe function:

mat = hp.matrix_from_dataframe(df, Winner = 'winner', Loser = 'loser')

Resulting in the following matrix of interactions:

We can get David's ranks and scores from such a matrix using the function david_ranks:

david = hp.david_ranks(mat)

And david will now be a dataframe with scores and ranks for each individual:

We can get the ADAGIO graph's nodes and edges from a matrix using the function run_ADAGIO. This function can take the arguments preprocess_data (Boolean, see paper for details), and plot (Also Boolean, whether or not to plot the resulting graph).

nodes, edges = hp.run_ADAGIO(mat, preprocess_data = False ,plot=True)

This results in the following plot:

From these nodes and edges, we can convert to rankings using the function rank_from_graph. We can use the argument method to choose "bottom-up" or "top-down" rankings (see paper for details):