测量点和组之间的距离

Question

I am trying to measure the distance between points inside a pandas dataframe. I first and looking to measure the distance between points that are in a sub region and get the average distance for that group. Then I want to measure the distance between the subregions (measuring the distance between those two vectors). I understand how to do the measuring part (using scipy.spatial.distance.euclidean for the former and scipy.spatial.distance.cdist for the latter). The issue I am running across is figuring out how to apply the functions to the dataset. I think I should use groupby.apply() and feed in my function, but I'm having trouble conceptualizing that. The dataframe looks like this:

id, latitude, longitude, subregion, region

Currently I have:

import pandas as pd
import numpy as np
from scipy.spatial.distance import euclidean

df = pd.read_csv('targets.csv')
...
def calculate_distance(x,y):
    return x._get_numeric_data().apply(axis=0, func=euclidean[x,y]).mean()

df.groupby('subregion').apply(calculate_distance)

I know this is incorrect as I want to apply to multiple columns for all the rows. My other thought is that I am using the wrong data structure for this.

Answer 1

I ended up using a different data structure and in the end looks like this:

contacts = {}

for i, row in sc_walkbook.iterrows():
    if contacts.get(row['region'],0) == 0:
        contacts[row['region']] = {}
        contacts[row['region']][row['subregion']] = {}
        contacts[row['region']][row['subregion']]['coords'] = []
        contacts[row['region']][row['subregion']]['distances'] = []
    elif contacts[row['region']].get(row['subregion'],0) == 0:
        contacts[row['region']][row['subregion']] = {}
        contacts[row['region']][row['subregion']]['coords'] = []
        contacts[row['region']][row['subregion']]['distances'] = []
    else:
        pass
    contacts[row['region']][row['subregion']]['coords'].append([row['T_Latitude'],row['T_Longitude']])

for region in contacts.itervalues():
    for subregion in region.itervalues():
        for a, b in itertools.combinations(subregion['coords'], 2):
            subregion['distances'].append(euclidean(a, b))