Cluster Center in Spark Streaming k-means Clustering

Asked Dec 27 '15 at 23:54

Active Dec 27 '15 at 23:54

Viewed 1,290 times

I am using Streaming k-means to cluster some 2-dimensional stream data using the example in

http://spark.apache.org/docs/latest/mllib-clustering.html#streaming-k-means.

code:

model = StreamingKMeans(k=5, decayFactor=0.7).setRandomCenters(2, 1.0, 0)
model.trainOn(trainingData)
clust=model.predictOnValues(testData.map(lambda lp: (lp.label, lp.features)))

It is working well without error. Now, I need to find and print the cluster center in each batch or over each sliding batch. Considering that the centroids are made with decayFactor of 0.7, how can I find/calculate the cluster centers?

asked Dec 27 '15 at 23:54

Saeed

1

Once the model is trained you can access the cluster centers via the method `clusterCenters()` – Christian Hirsch Dec 28 '15 at 08:01
AttributeError: 'TransformedDStream' object has no attribute 'clusterCenters' I think streaming kmeans doesnt have any clusterCenters() method. – Saeed Dec 28 '15 at 17:08
Can you give me the output of `model.latestModel().clusterCenters`? – Christian Hirsch Dec 29 '15 at 01:24
Yes, it is working, model.latestModel().clusterCenters shows the clusters. Thanks. – Saeed Dec 29 '15 at 02:14

Cluster Center in Spark Streaming k-means Clustering

0 Answers0