Compute LIME component error message

KNIME Extensions
1

Hi there,

Yesterday, Dec, 4th, I drag and dropped from Compute Local Model-agnostic Explanations (LIMEs) – KNIME Hub the Compute LIME componente into a workflow of mine, which was built in a very similar way to the example shown at that site.

When I did connect the inputs of the component, however, I got the following message: Execute failed: Explain Prediction Column 4:0:0:39: Train Local Linear Model 4:0:0:39:39: Compute Weighted RMSE 4:0:0:39:39:45: Contains an unconnected node (“H2O Predictor (Regression) 4:0:0:39:39:45:31”)

Is there any operation the user has to make internally at the component? I would very much appreciate your help about this issue.

2

Hi @JScheidegger Do you have the H2O extension installed?. Looks like the predictor H20 is giving you trouble. Open the component and check if the conexion is OK.

3

Hi @iperez, thank you for your answer. I do have the H2O extension installed, while, sure, I may have some part missing. When I open the component, over and over until something unconnected shows up, I find a metanode named Compute Weighted RMSE which has two inputs without any incoming line. This metanode has a H2O Predictor (Regression) at the very start that receives both open connections.