How to adjust this DBSCAN algorithm python

Question

I am using this clustering algorithm to cluster lat and lon points. I am using pre-written code which is given at http://scikit-learn.org/stable/auto_examples/cluster/plot_dbscan.html.

The code is as follows and takes in my file with over 4000 lat and lon points. However I want to adjust this code so that it only defines a cluster as points within say 0.000020 of each other, as I want my clusters to be almost at street level.

At the moment I am getting 11 clusters whereas in theory I want at least 100 clusters.I have tried adjusting and changing different figures but to no avail.

print(__doc__)

import numpy as np

from sklearn.cluster import DBSCAN
from sklearn import metrics
from sklearn.datasets.samples_generator import make_blobs
from sklearn.preprocessing import StandardScaler


##############################################################################
# Generate sample data
input = np.genfromtxt(open("dataset_import_noaddress.csv","rb"),delimiter=",", skip_header=1)
coordinates = np.delete(input, [0,1], 1)

X, labels_true = make_blobs(n_samples=4000, centers=coordinates, cluster_std=0.0000005,
                        random_state=0)

X = StandardScaler().fit_transform(X)

##############################################################################
# Compute DBSCAN
db = DBSCAN(eps=0.3, min_samples=10).fit(X)
core_samples_mask = np.zeros_like(db.labels_, dtype=bool)
core_samples_mask[db.core_sample_indices_] = True
labels = db.labels_

# Number of clusters in labels, ignoring noise if present.
n_clusters_ = len(set(labels)) - (1 if -1 in labels else 0)

print('Estimated number of clusters: %d' % n_clusters_)
print("Homogeneity: %0.3f" % metrics.homogeneity_score(labels_true, labels))
print("Completeness: %0.3f" % metrics.completeness_score(labels_true, labels))
print("V-measure: %0.3f" % metrics.v_measure_score(labels_true, labels))
print("Adjusted Rand Index: %0.3f"
  % metrics.adjusted_rand_score(labels_true, labels))
print("Adjusted Mutual Information: %0.3f"
  % metrics.adjusted_mutual_info_score(labels_true, labels))
print("Silhouette Coefficient: %0.3f"
  % metrics.silhouette_score(X, labels))

##############################################################################
# Plot result
import matplotlib.pyplot as plt

# Black removed and is used for noise instead.
unique_labels = set(labels)
colors = plt.cm.Spectral(np.linspace(0, 1, len(unique_labels)))
for k, col in zip(unique_labels, colors):
    if k == -1:
        # Black used for noise.
        col = 'k'

    class_member_mask = (labels == k)

    xy = X[class_member_mask & core_samples_mask]
    plt.plot(xy[:, 0], xy[:, 1], 'o', markerfacecolor=col,
         markeredgecolor='k', markersize=14)

    xy = X[class_member_mask & ~core_samples_mask]
    plt.plot(xy[:, 0], xy[:, 1], 'o', markerfacecolor=col,
         markeredgecolor='k', markersize=6)

plt.title('Estimated number of clusters: %d' % n_clusters_)
plt.show()

Has QUIT--Anony-Mousse · Accepted Answer · 2016-03-13 23:10:04Z

2

You appear to be changing the data generation only:

X, labels_true = make_blobs(n_samples=4000, centers=coordinates, cluster_std=0.0000005,
                    random_state=0)

instead of the clustering algorithm:

db = DBSCAN(eps=0.3, min_samples=10).fit(X)
            ^^^^^^^ almost your complete data set?

For geographic data, make sure to use haversine distance instead of Euclidean distance. Earth is more like a sphere than a flat Euclidean world.

answered Mar 13, 2016 at 23:10

Has QUIT--Anony-Mousse

77.8k14 gold badges146 silver badges198 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

RyanKilkelly Over a year ago

Thanks for your help, I have changed the eps=0.000010 and I am only getting 28 clusters whereas I was expecting alot more. Have you any suggestions as to why this is.Thanks!

Has QUIT--Anony-Mousse Over a year ago

Well, study the data. Adjacent blobs will be merged by DBSCAN. epsilon is not a maximum distance.

kendriu · Accepted Answer · 2023-09-23 06:58:23Z

0

Code on python, cluster image

from sklearn.cluster import DBSCAN
import numpy as np

# загрузка изображения
img = plt.imread('image.jpg')

# преобразование изображения в массив точек
points = np.reshape(img,(img.shape[0]*img.shape[1], img.shape[2]))

# определение алгоритма DBSCAN с параметрами
dbscan = DBSCAN(eps=10, min_samples=10)

# выполнение кластеризации
labels = dbscan.fit_predict(points)

# преобразование меток кластеров в изображение
clustered_img = np.reshape(labels, (img.shape[0], img.shape[1]))

# вывод изображения
plt.imshow(clustered_img)
plt.show()

edited Sep 23, 2023 at 6:58

kendriu

6375 silver badges22 bronze badges

answered Sep 21, 2023 at 4:41

igornet0

12 bronze badges

Collectives™ on Stack Overflow

How to adjust this DBSCAN algorithm python

2 Answers 2

2 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related