Transforming data for kmeans and PCA

jxn Published at Dev

jxn

I have a dataset that looks like this:

search_term = ['computer','usb port', 'phone adaptor']
clicks = [3,2,1]
bounce = [0,0,2]
conversion = [4,1,0]

I want to feed it into a kmeans model however i am having trouble transforming the lists into a matrix format so that it can be ingested by kmeans. I also want to reduce the dimensions with PCA so it can be visualized in a 2d plot.

This is what my code looks like:

X = np.array(clicks, bounce, conversion)
y = np.array(search_terms)
num_clusters = 3

pca = PCA(n_components=2, whiten=True).fit(X)
X_pca = pca.transform(X)

km=KMeans(n_clusters=num_clusters, init='k-means++',n_init=10, verbose=1)
km.fit(X_pca)

print km.labels_[:10]

This is the error i got:

TypeError: unsupported operand type(s) for -: 'numpy.ndarray' and 'numpy.ndarray'

Also, once the clustering is done, i would hope to be able to see which search terms fall into which cluster so i'm not sure if setting y = np.array(search_terms) is correct?

Please advise.

Jianxun Li

The following code should work. Let me know if this is not the case.

import numpy as np
from sklearn.decomposition import PCA
from sklearn.cluster import KMeans

search_terms = ['computer','usb port', 'phone adaptor']
clicks = [3,2,1]
bounce = [0,0,2]
conversion = [4,1,0]

X = np.array([clicks, bounce, conversion]).T
y = np.array(search_terms)

num_clusters = 3

X_pca = PCA(n_components=2, whiten=True).fit_transform(X)

km = KMeans(n_clusters=num_clusters, init='k-means++',n_init=10, verbose=1)
km.fit(X_pca)

Collected from the Internet

Please contact [email protected] to delete if infringement.