How to create clusters using DBSCAN in Python

Density Based Spatial Clustering of Applications with Noise(DBSCAN) is one of the clustering algorithms which can find clusters in noisy data. It works even on those datasets where K-Means fail to find meaningful clusters. More information about it can be found here.

You can learn more about the DBSCAN algorithm in the below video.

The below code snippet will help to create clusters in data using DBSCAN.

Creating data for clustering

# importing plotting library 
import matplotlib.pyplot as plt
# Create Sample data
from sklearn.datasets import make_moons
X, y= make_moons(n_samples=500, shuffle=True, noise=0.1, random_state=20)
plt.scatter(x= X[:,0], y= X[:,1])

# importing plotting library

import matplotlib.pyplot as plt

# Create Sample data

from sklearn.datasets import make_moons

X, y= make_moons(n_samples=500, shuffle=True, noise=0.1, random_state=20)

plt.scatter(x= X[:,0], y= X[:,1])

Sample Output:

Finding Best hyperparameters for DBSCAN using Silhouette Coefficient

The Silhouette Coefficient is calculated using the mean intra-cluster distance (a) and the mean nearest-cluster distance (b) for each sample. The Silhouette Coefficient for a sample is (b – a) / max(a, b). To clarify, b is the distance between a sample and the nearest cluster that the sample is not a part of. Note that Silhouette Coefficient is only defined if number of labels is 2 <= n_labels <= n_samples – 1.

The best value of the Silhouette Coefficient is 1 and the worst value is -1. Values near 0 indicate overlapping clusters. Negative values generally indicate that a sample has been assigned to the wrong cluster

## Finding best values of eps and min_samples
import numpy as np
import pandas as pd
from sklearn.metrics import silhouette_score
from sklearn.cluster import DBSCAN

# Defining the list of hyperparameters to try
eps_list=np.arange(start=0.1, stop=0.9, step=0.01)
min_sample_list=np.arange(start=2, stop=5, step=1)

# Creating empty data frame to store the silhouette scores for each trials
silhouette_scores_data=pd.DataFrame()

for eps_trial in eps_list:
    for min_sample_trial in min_sample_list:
        
        # Generating DBSAN clusters
        db = DBSCAN(eps=eps_trial, min_samples=min_sample_trial)
        
        if(len(np.unique(db.fit_predict(X)))&gt;1):
            sil_score=silhouette_score(X, db.fit_predict(X))
        else:
            continue
        trial_parameters="eps:" + str(eps_trial.round(1)) +" min_sample :" + str(min_sample_trial)
        
        silhouette_scores_data=silhouette_scores_data.append(pd.DataFrame(data=[[sil_score,trial_parameters]], columns=["score", "parameters"]))

# Finding out the best hyperparameters with highest Score
silhouette_scores_data.sort_values(by='score', ascending=False).head(1)

## Finding best values of eps and min_samples

import numpy as np

import pandas as pd

from sklearn.metrics import silhouette_score

from sklearn.cluster import DBSCAN

# Defining the list of hyperparameters to try

eps_list=np.arange(start=0.1, stop=0.9, step=0.01)

min_sample_list=np.arange(start=2, stop=5, step=1)

# Creating empty data frame to store the silhouette scores for each trials

silhouette_scores_data=pd.DataFrame()

for eps_trial in eps_list:

for min_sample_trial in min_sample_list:

# Generating DBSAN clusters

db = DBSCAN(eps=eps_trial, min_samples=min_sample_trial)

if(len(np.unique(db.fit_predict(X)))>1):

sil_score=silhouette_score(X, db.fit_predict(X))

else:

continue

trial_parameters="eps:" + str(eps_trial.round(1)) +" min_sample :" + str(min_sample_trial)

silhouette_scores_data=silhouette_scores_data.append(pd.DataFrame(data=[[sil_score,trial_parameters]], columns=["score", "parameters"]))

# Finding out the best hyperparameters with highest Score

silhouette_scores_data.sort_values(by='score', ascending=False).head(1)

Sample Output

Creating clusters using the best hyperparameters

# DBSCAN Clustering
from sklearn.cluster import DBSCAN
db = DBSCAN(eps=0.18, min_samples=2)
# Plotting the clusters
plt.scatter(x= X[:,0], y= X[:,1], c=db.fit_predict(X))

# DBSCAN Clustering

from sklearn.cluster import DBSCAN

db = DBSCAN(eps=0.18, min_samples=2)

# Plotting the clusters

plt.scatter(x= X[:,0], y= X[:,1], c=db.fit_predict(X))

DBCAN clustering in python — DBSCAN clustering in python

Author Details

Farukh Hashmi

Lead Data Scientist

Farukh is an innovator in solving industry problems using Artificial intelligence. His expertise is backed with 10 years of industry experience. Being a senior data scientist he is responsible for designing the AI/ML solution to provide maximum gains for the clients. As a thought leader, his focus is on solving the key business problems of the CPG Industry. He has worked across different domains like Telecom, Insurance, and Logistics. He has worked with global tech leaders including Infosys, IBM, and Persistent systems. His passion to teach inspired him to create this website!

https://thinkingneuron.com/

thinkingneuron@gmail.com

2 thoughts on “How to create clusters using DBSCAN in Python”

Rebecca V.
August 2, 2022 at 6:02 pm

Hi! Thanks for the code snippet. Just a heads up it appears there may be a rendering error in line 20:

if(len(np.unique(db.fit_predict(X)))>1):

1. Rebecca V.
  August 2, 2022 at 6:03 pm
  
  Of course it’s rendering properly in my comment lololol. Anyway, thanks again!

Creating data for clustering

Finding Best hyperparameters for DBSCAN using Silhouette Coefficient

Creating clusters using the best hyperparameters

2 thoughts on “How to create clusters using DBSCAN in Python”

Leave a Reply! Cancel Reply