Leer Defining the Number of Clusters

There are several techniques we can use to help us to define the optimal number of clusters. There are also several techniques available for spectral clustering, but they are strongly based on hard math and are not implemented within several functions.

What can we use there? Remember the second section and the silhouette score we considered. We can also use it there.

Let's see what will be the result of building the silhouette scores chart for the circles' data (the scatter plot is below).


              12345678910111213141516171819202122232425
            
# Import the libraries
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.cluster import SpectralClustering
from sklearn.metrics import silhouette_score

# Read the data
data = pd.read_csv('https://codefinity-content-media.s3.eu-west-1.amazonaws.com/138ab9ad-aa37-4310-873f-0f62abafb038/model_data4.csv', index_col = 0)

# Creating lists
n_cl = range(2, 10)
silhouettes = []

# Calculate the scores for different number of clusters
for j in n_cl:
    model = SpectralClustering(n_clusters = j, affinity = 'nearest_neighbors')
    model.fit(data)
    silhouettes.append(silhouette_score(data, model.labels_))
    
# Visualize the results
g = sns.lineplot(x = n_cl, y = silhouettes)
g.set_xlabel('Number of clusters')
g.set_ylabel('Silhouette score')
plt.show()

Taak

Swipe to start coding

Build the silhouette score chart for the weather data using silhouette scores and spectral clustering. Follow the next steps:

Import SpectralClustering and silhouette_score functions from sklearn.cluster and sklearn.metrics respectively.
Create a range object named n_cl with integer numbers from 2 to 9 (inclusive).
Iterate over n_cl. On each step:

Create SpectralClustering model named model with j clusters and 'nearest_neighbors' affinity.
Fit (.fit() method) the numerical columns of data to model. The numerical columns are 3 - 14.
Append to silhouettes list the value of silhouette score. Pass the predicted labels_ as the second parameter.

Build the seaborn line plot n_cl (x-axis) vs silhouettes (y-axis)

Oplossing

Was alles duidelijk?

Bedankt voor je feedback!

Sectie 4. Hoofdstuk 4

single

Vraag AI

Vraag wat u wilt of probeer een van de voorgestelde vragen om onze chat te starten.

Awesome!

Completion rate improved to 3.57

Veeg om het menu te tonen