determine_optimal_elbow.py

import pandas as pd


def determine_optimal_elbow(vals):
    elbow_data = pd.DataFrame(columns=['index', 'k', 'delta'])
    last = None
    i = 0
    for x in vals:
        if (last is None):
            last = x
        delta = last - x
        row = [i, x, delta]
        elbow_data.loc[i] = row
        i += 1
        last = x
    # we want the cluster AFTER the largest delta, unless it is the last cluster in the array
    row = elbow_data.loc[elbow_data["delta"].idxmax()]
    row_count = 0
    if row["index"] < len(elbow_data) - 1:
        row = elbow_data.loc[row["index"] + 1]
    return row


vals = [9.120197530868204,
        6.084063492055672,
        2.0175925925981977,
        1.3755925925991808]

print(determine_optimal_elbow(vals))