Agrupamento por distâncias - VMD.ipynb

Kernel: Python 3 (Ubuntu Linux)

from scipy.cluster.hierarchy import dendrogram, linkage, fcluster
from scipy.spatial.distance import pdist, squareform
from matplotlib import pyplot as plt
from geopy.distance import vincenty
import geopy
import numpy as np
import pandas as pd

max_d = 2 # Distancia maxima de 2km]

def calcula_distancias(locs_1, locs_2):
    n_linhas_1 = locs_1.shape[0]
    n_linhas_2 = locs_2.shape[0]
    dists = np.empty((n_linhas_1, n_linhas_2))
    for i in range(n_linhas_1):
        for j in range(n_linhas_2):
            dists[i, j] = geopy.distance.vincenty(locs_1[i], locs_2[j]).km
    return dists

locais = pd.read_excel('Planilha para distribuicao do efetivo.xlsx', sheet_name='dados')
locais[:5]

coords = locais[['longitude','latitude']].values

coords

array([[-61.50422603, -10.87078709],
       [-60.941759  , -10.475339  ],
       [-60.180703  , -14.107886  ],
       ...,
       [-50.511945  , -10.47179925],
       [-50.51073158, -10.46768262],
       [-50.296635  ,  -9.994842  ]])

matriz_distancias = calcula_distancias(coords, coords)

matriz_distancias

array([[0.00000000e+00, 6.61801777e+01, 2.29587452e+02, ...,
        1.22413550e+03, 1.22427559e+03, 1.24901215e+03],
       [6.61801777e+01, 0.00000000e+00, 2.16509550e+02, ...,
        1.16121113e+03, 1.16134618e+03, 1.18553983e+03],
       [2.29587452e+02, 2.16509550e+02, 0.00000000e+00, ...,
        1.10042873e+03, 1.10061520e+03, 1.13048953e+03],
       ...,
       [1.22413550e+03, 1.16121113e+03, 1.10042873e+03, ...,
        0.00000000e+00, 3.21690889e-01, 4.15135648e+01],
       [1.22427559e+03, 1.16134618e+03, 1.10061520e+03, ...,
        3.21690889e-01, 0.00000000e+00, 4.11970411e+01],
       [1.24901215e+03, 1.18553983e+03, 1.13048953e+03, ...,
        4.15135648e+01, 4.11970411e+01, 0.00000000e+00]])

d = matriz_distancias.shape[0]

v = d*(d - 1)/2
C = np.empty([int(round(v,0))])

ic = 0
for i in range(d):
    for j in range(i+1,d):
        # print(matriz_distancias[i,j])
        C[ic] = matriz_distancias[i,j]
        ic  =  ic + 1

C

array([ 66.18017766, 229.587452  , 283.50213756, ...,   0.32169089,
        41.5135648 ,  41.19704105])

Z = linkage(C, 'complete')
dn = dendrogram(Z)

# https://docs.scipy.org/doc/scipy/reference/generated/scipy.cluster.hierarchy.linkage.html

grupos = fcluster(Z, max_d, criterion='distance')

locais['GRUPOS'] = grupos

locais[:5]

res1 = pd.ExcelWriter('res1 VMD.xlsx')
locais.to_excel(res1,'dados',index=False)
res1.save()