Tree Map not working in error analysis

[karthi0804]

Describe the bug I couldnt get the tree map in Error analysis.

To Reproduce I followed the steps in the notebook

Screenshots

Desktop (please complete the following information):

• OS: Ubuntu 18.04
• Browser chrome
• Python version: 3.8
• raiwidgets and responsibleai package versions [e.g. 0.19.0]

To get the package versions please run in your command line:

Name: raiwidgets
Version: 0.26.0
Name: responsibleai
Version: 0.26.0

Please help if i am missing anything

[hawestra]

Hi @karthi0804, sorry for the delay in response; I wasn't able to repro this issue using the latest version of raiwidgets and responsibleai. Are there any errors in the console ? Are you running this locally or using a remote compute ?

[LucieNvz]

Hello, I am currently facing the same problem. Nothing is displayed in the error analysis component. The tree map is empty. I am running code remotely, and am using the latest versions of the required packages. I also followed the same notebook tutorial.

The other components seem to work as expected.

[WalidBenAhmed]

Hi I'm facing a similar problem, but in my case there is only one node in my tree

[imatiach-msft]

@WalidBenAhmed it looks like everything is working as intended for you - just every example has an error. Why specifically all true and predicted labels don't align is another question though - perhaps taking a screenshot of the table view might help us understand what is causing all instances to be invalid here.

[WalidBenAhmed]

@imatiach-msft, thank you for your reply I'm using the example responsibleaidashboard-diabetes-regression-model-debugging.ipynb (attached) you can use it to reproduce the issue responsibleaidashboard-diabetes-regression-model-debugging.zip

Method 1: doesn't work => we have one node rai_insights = RAIInsights(model, train_data, test_data, target_feature, 'regression', feature_metadata=feature_metadata) rai_insights.explainer.add() rai_insights.error_analysis.add()

Method 2, using the same data and model: works predictions = model.predict(X_test) ErrorAnalysisDashboard(dataset=X_test, true_y=y_test, features=continuous_features, pred_y=predictions, model_task='regression', max_depth=3)