from sklearn.cluster import KMeans

import numpy as np

# Load data

data = np.loadtxt('data.csv', delimiter=',')

# Initialize k-means model with desired number of clusters

kmeans = KMeans(n_clusters=4)

# Fit the model to the data

kmeans.fit(data)

# Calculate feature importances

importances = kmeans.feature_importances_

# Print the feature importances

print(importances)

After a K-mean clustering, for instance, to investigate children's food options selections patterns, I am checking if the cluster membership is associated with individual characteristics (e.g., group of age, economic statuses such as universal free school meal and free school meal means-tested entitlement) using chi-squared tests. K-means clustering is used to do a data exploration analysis. If you need to have variable importance, you have to use supervised machine learning of regression or classification dependent on your research question.