Watson Studio, Watson ML, Watson OpenScale

 View Only

  • 1.  K-means PMML model to Python

    Posted Fri December 16, 2022 09:57 AM
    I first created a K means model in Modeler and then I created a model in Python with Scikit learn library. So when I compared the results of both models they were different, the question is ¿is there anyway to use de PMML exportation of the model in Python? or ¿How can I replicate the model of K means from Moderler in Python?

    ------------------------------
    Eduardo García
    ------------------------------

    #SPSSModeler


  • 2.  RE: K-means PMML model to Python

    User Group Leader
    Posted Wed June 21, 2023 07:31 AM

    Here's an example of how you can export the K-means model from Modeler and import it into Python using pmml-sklearn:

    1. Export the K-means model from IBM SPSS Modeler as a PMML file.
    2. Install the pmml-sklearn library using pip install pmml-sklearn.
    3. Load the PMML file and use it to create the K-means model in Python.
    

    from sklearn_pmml import sklearn2pmml
from sklearn_pmml.convert import estimator_to_pmml
from sklearn.cluster import KMeans

# Load your data
# ...

# Create the K-means model
kmeans = KMeans(n_clusters=3, init='k-means++', random_state=0)

# Fit the model to your data
kmeans.fit(data)

# Export the K-means model to PMML
pmml_file = "kmeans_model.pmml"
sklearn2pmml(kmeans, pmml_file, with_repr=True)

# Import the K-means model from PMML
from sklearn_pmml import PMMLPipeline
pipeline = PMMLPipeline.from_pmml(pmml_file)

# Get the cluster labels for your data points
labels = pipeline.predict(data)

# Get the cluster centers
centers = pipeline.named_steps['kmeans'].cluster_centers_

    Hope this helps !



    ------------------------------
    Youssef Sbai Idrissi
    Software Engineer
    ------------------------------

    Original Message