H2O MOJO Predictor(Autoencoder/Dimension Reduction) nodes usage

[mihais1]

Hi,

I am having a problem using Spark H2O/H2O nodes: Predictor(Autoencoder) and Predictor(Dimension Reduction). I have warnings regarding to the MOJO category which must be DimReduction or AutoEncoder. I uploaded the modified workflow example: “kn_example_h2o_sparkling_water.knwf” from the site. The same problem is when I am using H2O k-Means node instead of H2O Random Forest Learner node.

And I have another question: how can I activate the Sparkling Water Flow UI? I tried to set the “spark.ext.h2o.node.enable.web” key TRUE for the Create H2O Sparkling Water Context node on another workflow on a Linux OS and I have this error: “\n\nERROR MESSAGE:\n\nResource/flow/index.html not found\n\n”. I uploaded the image with this error.

Thank you,
Mihai

kn_example_h2o_sparkling_water_autoencoder.knwf (110.4 KB)

[mlauber71]

Thank you for using my example with H2O.ai and Spark. You will have to connect the MOJO File (Reader) to a compatible model from H2O. At the moment they connect to a classification model.

As far as I can see MOJOs for Dimension reduction are not yet represented with individual KNIME/H2O nodes. It might be possible to import such model from R or Python implementations of H2O. I would have to investigate.

Combine Big Data, Spark and H2O.ai Sparkling Water – mlauber71

It should be possible to build a wrapper for KNIME like I demonstrated for H2O’s AutoML.

H2O.ai AutoML in KNIME for classification problems

H2O.ai Automl - a powerful auto-machine-learning framework wrapped with KNIME It features various models like Random Forest or XGBoost along with Deep Learning. It has wrappers for R and Python but also could be used from KNIME. The results will be written to a folder and the models will be stored in MOJO format to be used in KNIME (as well as on a Big Data cluster via Sparkling Water). One major parameter to set is the running time the model has to test various models and do some hyperparam…

Regarding your other question. If you check the context of the local big data node with right click it will provide you with a link. There is also a tab with Sparkling Water information. But unfortunately I was not immediately able to open the GUI.