Description
Currently, null/missing values from training datasets are displayed inconsistently across the web application. Since some models (e.g. CatBoost) natively support missing values, we should provide a unified representation throughout the UI to avoid confusion for users.
Current Behavior
Null values are displayed differently depending on the component:
- Contribution Plot
- Local Plot
- Dataset panel/table
- Contribution Plot hover text
- Cluster Plot hover text
- Local Plot hover text
Motivation
A consistent display of missing values improves readability and avoids ambiguity when exploring model inputs and explanations. This is particularly important for models that can directly handle missing values, where null entries are legitimate feature values rather than data quality issues.
Description
Currently, null/missing values from training datasets are displayed inconsistently across the web application. Since some models (e.g. CatBoost) natively support missing values, we should provide a unified representation throughout the UI to avoid confusion for users.
Current Behavior
Null values are displayed differently depending on the component:
Motivation
A consistent display of missing values improves readability and avoids ambiguity when exploring model inputs and explanations. This is particularly important for models that can directly handle missing values, where null entries are legitimate feature values rather than data quality issues.